01. PYMC - Bayesian ABCD question type.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pymc3 as pm\n",
    "from scipy import stats\n",
    "from custom_metrics import *\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "import pandas as pd\n",
    "from IPython.display import Image, display\n",
    "from scipy.special import gammaln ,digamma\n",
    "from tqdm import tqdm\n",
    "import sys"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h1><center>A Probabilistic Model for \"Yes\" or  \"No\" Queries:  \n",
    "Crowdsourcing Multi-Class Classification Tasks </center></h1> \n",
    "<h2><center>Compare to Standard Complete ABCD question type</center></h2>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**1. Select the folder data:**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "selected_folder = './catalina/'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**2. Load the data:**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# users + number of YN question selected for them as.\n",
    "info_experts = pd.read_csv(selected_folder+'users.csv') "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# responses for the YN queries\n",
    "votes = pd.read_csv(selected_folder+'votes_yn.csv') \n",
    "\n",
    "# objects to be classified\n",
    "objects = pd.read_csv(selected_folder+'objects.csv')\n",
    "# picking only object which were asked for both query types\n",
    "objects = objects[objects._abc | objects._training]\n",
    "\n",
    "# users + number of YN question selected for them as.\n",
    "info_experts = pd.read_csv(selected_folder+'users.csv') \n",
    "info_experts = info_experts.set_index('id')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training set size 41\n",
      "Testing set size 80\n",
      "Number of objecs required ABCD and YN votes 80\n"
     ]
    }
   ],
   "source": [
    "print(\"Training set size {}\".format(objects._training.sum()))\n",
    "print(\"Testing set size {}\".format(objects._validation.sum()))\n",
    "print(\"Number of objecs required ABCD and YN votes {}\".format(objects._abc.sum()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Selecting the users who complete at least 70% of their contest's questions:\n",
    "number_of_given_answers = dict(votes.expert_id.value_counts())\n",
    "info_experts['percentage_done'] = 0\n",
    "for key in number_of_given_answers:\n",
    "    _total_question_created_for_this_expert = float(info_experts.loc[key].initial_yn_questions)\n",
    "    info_experts.loc[key].percentage_done = float(number_of_given_answers[key])/_total_question_created_for_this_expert\n",
    "info_experts = info_experts.reset_index()\n",
    "info_experts = info_experts[info_experts.percentage_done >= .7]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**3 .Selecting data for running the model:**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Selected annotators\n",
    "experts = info_experts.id.values\n",
    "\n",
    "# Filtering the votes\n",
    "votes = votes[votes.expert_id.isin(experts)]\n",
    "\n",
    "o_columns = objects[['catalina_id','Var_Type','_training','_validation']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([19, 10,  9, 12,  6,  4, 13, 14]), 8)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "experts, experts.shape[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['CEP', 'EB', 'LPV', 'RRLYR'], dtype=object)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "o_columns.Var_Type.unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Number of possible classes\n",
    "K = o_columns.Var_Type.unique().shape[0]\n",
    "\n",
    "# Number of annotators\n",
    "J = experts.shape[0]\n",
    "\n",
    "# Fit the annotators id with a sequential number\n",
    "inv_experts_id = dict(zip(experts, range(J)))\n",
    "\n",
    "# Fit the classes names with a sequential number\n",
    "classes_dict = dict(zip(range(K),list(o_columns.Var_Type.unique())))\n",
    "inv_classes_dict = {classes_dict[a]:a for a in classes_dict}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# I. Data Training Credibility Stage"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Getting the true classes' dirstribution\n",
    "Ns_dict = dict(o_columns[o_columns._training].Var_Type.value_counts())\n",
    "Ns = np.array([Ns_dict[classes_dict[i]] for i in classes_dict])\n",
    "N = np.sum(Ns)\n",
    "\n",
    "# Getting the true proportion classes' dirstribution\n",
    "p = Ns / N\n",
    "\n",
    "# LABELS GROUND TRUTH - TRAINING SET\n",
    "zt = np.array([inv_classes_dict[_] for _ in o_columns[o_columns._training].Var_Type])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Generatting the r matrix of votes from the source data\n",
    "def matrix_votes(J, N, M, o_columns, votes):\n",
    "    r_ = np.zeros((J, N, M, 2))\n",
    "    for i in range(o_columns.shape[0]):\n",
    "        instance = o_columns.iloc[i]\n",
    "        _votes = votes[votes.object_id == instance.catalina_id]\n",
    "        for v in range(_votes.shape[0]):\n",
    "            vote = _votes.iloc[v]\n",
    "            r_[inv_experts_id[vote.expert_id], i,  inv_classes_dict[vote.question], 1-vote.value] = 1\n",
    "    return r_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Getting the votes for the users KNOWN LABELS - TRAINING PART\n",
    "r_true = matrix_votes(J, N, K, o_columns[o_columns._training], votes)  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# II. Data Testing Labeling Stage"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Getting the true classes' dirstribution\n",
    "Ns_test_dict = dict(o_columns[~o_columns._training].Var_Type.value_counts())\n",
    "Ns_test = np.array([Ns_test_dict[classes_dict[i]] for i in classes_dict])\n",
    "N_test = Ns_test.sum()\n",
    "\n",
    "# Getting the true proportion classes' dirstribution\n",
    "p_test = Ns_test/N_test\n",
    "\n",
    "# LABELS GROUND TRUTH - TESTING SET\n",
    "zt_test = np.array([inv_classes_dict[_] for _ in o_columns[~o_columns._training].Var_Type])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Getting the votes for the users UKNOWN LABELS - TESTING PART\n",
    "r_test = matrix_votes(J, N_test, K, o_columns[~o_columns._training], votes)  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# III. YN - PYMC IMPLEMENTATION"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### III. a) YN implementation - Credibility"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 20000/20000 [00:06<00:00, 3002.91it/s]\n"
     ]
    }
   ],
   "source": [
    "ALPHA = .5\n",
    "BETA = .5\n",
    "\n",
    "with pm.Model() as model: \n",
    "    \n",
    "    theta = pm.Beta('theta', alpha=ALPHA, beta=BETA, shape=(J, K, K))\n",
    "    z = pm.Categorical('z', p=p , observed=zt, shape=N)\n",
    "\n",
    "    # EQUATION (4.3)\n",
    "    def likelihood_Pt_rz(value, theta=theta, zt=z):\n",
    "        l=np.log(theta[:, zt,:]**value[:,:,:,0] * (1-theta[:,zt,:])**(value[:,:,:,1]))\n",
    "        return l.sum()\n",
    "    \n",
    "    r = pm.DensityDist('r', likelihood_Pt_rz, observed=(r_true))\n",
    "    \n",
    "with model:\n",
    "    step = pm.Metropolis([theta])\n",
    "    trace = pm.sample(20000, step)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[<matplotlib.axes._subplots.AxesSubplot object at 0x7fad6015b780>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fad553530f0>]], dtype=object)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA1gAAACICAYAAAD6SUoaAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzs3VmsZFue3/Xvf609xnzGzDyZN2/eoe6toavbXSrcbkHb\nSPCAEZYBWS0QtmQkZFkyPPCAhGQE+MH4DclSAy1LDAIZjB9AGGFZgLAoeqK6q7qrq7qGO+Z4Ms8Y\n4449rbX+PERWUWr3UN2ddbNv3fWRjuJE7H12/CNin732L/bea4mqEkVRFEVRFEVRFP3RmZddQBRF\nURRFURRF0Y+KGLCiKIqiKIqiKIpekBiwoiiKoiiKoiiKXpAYsKIoiqIoiqIoil6QGLCiKIqiKIqi\nKIpekBiwoiiKoiiKoiiKXpAYsKLoh0hE7omIikjysmuJoiiKoh9EbLui6I8mBqwoesFE5L6I/PMv\nYDmxgYuiKIo+ErHtiqIXJwasKIqiKIqiKIqiFyQGrCh6gUTkvwPuAv+riGyAn30+6d8QkYciciki\nf/375jci8u+LyPsiciUif19E9p9P/tLz24WIbETkp0XkDRH5v57Peykif1dEZh/dK4yiKIp+1MS2\nK4perBiwougFUtW/BDwE/pyqjoC//3zSPwO8DfxzwH8oIp95/vi/A/zLwJ8BToA58J89n/ann9/O\nVHWkqr8MCPC3ns/7GeAV4D/+Yb6mKIqi6EdbbLui6MWKASuKPhp/Q1VrVf0a8DXgJ54//leBv66q\nj1W1Zdfg/IXf7dx1VX1PVf8PVW1V9QL4T9k1cFEURVH0osW2K4r+EOIFiFH00Xj2fb9vgdHz318F\n/mcRCd833QM3fqeFiMgN4G8DPwOM2X1JMn/h1UZRFEVRbLui6A8lHsGKohdP/wDzPgL+rKrOvu+n\nUNUnv8ty/pPnj39eVSfAX2R36kUURVEU/VHEtiuKXpAYsKLoxTsDXv8B5/154G+KyKsAInIkIn/+\n+bQLIPy2ZY2BDbAUkdvAv/diSo6iKIo+4WLbFUUvSAxYUfTi/S3gPxCRBfAXfp95/zbwD4D/XUTW\nwK8APwWgqlvgbwK/KCILEflTwN8AvgAsgf8N+J9+OC8hiqIo+oSJbVcUvSCi+gc5IhxFURRFURRF\nURT9buIRrCiKoiiKoiiKohckBqwoiqIoiqIoiqIXJAasKIqiKIqiKIqiFyQGrCiKoiiKoiiKohfk\nYzHQ8OHhod67d+9llxFFURS9IF/5ylcuVfXoZdfxwxTbriiKoh8tP2jb9bEIWPfu3ePXfu3XXnYZ\nURRF0QsiIg9edg3fJSL/FfAvAeeq+mO/w3Rh1y31vwhsgb+sql/9/ZYb264oiqIfLT9o2xVPEYyi\nKIo+6f4b4F/4Pab/WeBTz3/+CvBffAQ1RVEURR9TH4sjWFEURdEfD/Nnp8xu3GJ3UOdHg6p+SUTu\n/R6z/Hngv9XdwJG/IiIzEbmlqk8/kgL/kFQVFxxd6PDq8cHj1WPFIipYY0GgTEoSk3zvb3CKugCq\nqIL3gU3jMAKpNRgjZNbQ9x0u9JgANk2xSYrD0/seYwyZybDG4NUjAdq6BqDXDi9KYXKMNYCA7J7b\nmP//PoB6paejDoGmWtPOl4gkpLZEEoNYg7EWRJDnNSMCRhAjKCAiSAiQJIiAYHavU4SAgrV45wlt\ngw0tEgAE7wNdcEhqsFiMGtI8wSQZAGINIub5e+vIswFbcZBZvjvGqJHd6xEEEWC7QdRgshzyktxm\nWAHaFYLsahcDYhBjUJMixu6e77d9vuK2oAFRBe++97diDNgUxYAKeZE9f19BQ9jdOofJsn9ypQmB\n0PdIlrHtPF3bYZqaNLH4oNiywKnCtsIaAyIk+QBJLWmW/K7bBVWFvt89//N5xBgkTQneY6wltBWm\nr8AkoAENDt/3tJs12naQZSAGU5S7hZoM0gKxCSioBtIsJ8kyFHBdoHcBUbBWyMrfVp8qBE+Poe49\nAjivWAO5ekwIJMMBIkIIAWMMbdsSvvseBuV61aGqDHMLIdDUNX1QrBjSxFIOyu/7wOR7n6NqIC2S\n730u3jnW9RJVZZCWWJvs1mkEMQab/Lbdda9gdv8noXFULuCNEFQxIlgRvCpFajEGUmvpfaDtA30I\nz+eBXGvwnmwwQVDWm4quqUm1YzQc0pOgfU2nPWIMhcnpupYssRgB4w0msQjCstrSB6XMUwZFTu8c\nddMAhuBaAGyRYQIoEIIjGwxJhnsEVdrthizJsTZn23RYY1FjaZ2j9x5jDOoDvnckRhkOhyTWUPcB\nHwJOFUFIMkNilGXfMEwsB6miCF4S1LUU4/3fcR190T4RAWu1+k2SZMxg8NrLLiWKouhjq96s+a//\n3b/KP/2zf5Gf+ld+9mWX81G6DTz6vvuPnz/2TwQsEfkr7I5ycffu3Y+kOIAQAnVdk6YpSZLwYP2A\nb11963vTz+tzTtenuHXL+VVg28Od+gP+qe4nSEJOl9dweMWy8pxtOqb1Ieut52l5g0XyhMKX5IMp\nYbMhD7udRqvCcV8yMqdIeERgFya8GlrNKbOEbO+Qo+4J53VBoh2aJQSXo3aCygRb1IS2YFsZBHjq\n5kzbIa8tPGsWdP0K1NAZg01KjBZULiWnZi8ZkptdLTlnjMNTnpU38LILJAGHywqMdqRhS5JOyXxK\n7faev2ceQbChJnPb771XSbNiER6xBHJTo77kWlKqbAPAVNZMZYO4kj5RVG6hKGVZECTHi2Fdjlne\nOSbLLW3jOX52je3ndNsaJymtSbECoz4jC8rlyU0+fXNM2Xvkw6cc5R4ROGsXXLuKJrnF4NWfJtt7\njcHVe6y6Lb3vOMg2lKOOwewIFwISHDZJsL/1AZerKWe+IYTASCbM8hLvDetQMxts8ZogNtDuH7J+\n/C7SLilMycYMGWQdPhEKTcjqlvOQoMUhdjDE5inaN6w3KV3nGYxGIDAuEm6mlqvgqbsPSK+uGBxO\nCNmM/irwuGrZTxxFV4FNae2AroFneUoyU+64Z2TbQD4suJs5BmdzGttSDkagwsZPGYQFq23Js+qC\no2HK8jOvMc9vcDbPWbiMJhg2Xcvnbc4ZW8yw5IQpR5sFpn4P7RP2q8fI3k2q5jH5eIB89i264PiW\nvcedL/86o+6MkChXXU6XzUgmB1yfnVLsDdkcfAqVhO16RVaMkKBk7Qq/qnieqXfrkBVSA4nbUB9n\nMB5hijHTtsCIofENV+2clVuT2FOmmZBkFg0Wg2Oy91mOyJHNBWfFADEZZTnhFZdj6idskyNOuz1s\naBjW7/G0SzjVGzTbFTfSSz6XbtHQsti/RcURkz7QOE9zfIPSX1M/fg8fLHVlMcHjsbhkQJCColsQ\n0h7f1TiUZuAobEeWJ2TWIOGYy02G74UiQJUqQ3Ec5AbtWzR4xBimoyHXy2vsokNE2fY9d8I5vRlT\nJ4eItLTaU0iPHNwmMx4XhMob0ECwOcF7akpS7RBpmJoS9QGTJmBhOhzybhoIZkAgh+W7BCc06iCA\nWMgPKvas4SZDTsobfPFf+7fA/PBP4PuRD1jOVXz11/8SB/s/w+c//3Mvu5woiqKPrQ++8mU0BF79\n8Z982aX8saWqfwf4OwBf/OIX9feZ/Y/k2bKhzCxFavjv/9EvAPCFu3s8qR/z1euvcmN0hzdfuUeZ\nHPKl9+9zp52xelaRX+Z8Rn6DKffw6wFd58j6nKviim8fWNRA+vg73KpbdPKA7s5NJj345oJu0JJ3\nCfnSUKRDJIOsz1glGdo4RIQnQ4VQc7d32PNrni5XeIG0vMFgfIM+KUjNEJpzlnqTbZqymW7ZrJ4h\n1Ya2veRUhuAqMAaDobU96dEBT6Wk7HOcLqhtyadHUyYOrrcptU2hzNk7uMf1ak6oKoJYQm9wAsP6\njC67w3CUkhzs8371EKmvOJhf0wwHXFPh6gP8/md5Uu5xXd5nz4xZdSki0JMzCw34MVV3C+Mdff4h\nr9QrNoOCzFSMXUZd/jS3mprj+xvmdY1lj1ZnZN4hsylbLwyqLZ3ZshSHMxCePeCR/xRTNUyGt3mA\nMrRbrvyKsA649gHB3eAoechynOBRbJJw4W9y4A549z6E1tAe17y+2LA5nbFOVmhTob6l1XM2icGb\ngk024Vkx5CZHbNorBg/ex/dzRKAKFRIqHmtH5lLKJGE19mgIGC7o+wnl0zF29CZXrmcmLUVyhnY5\nm6Xl0dUzFn7IOt1ixTB5+ISh/ZBTv8eoVNZJxoWdMNaKbjtnFUrSLrB39oiKQFXuY9YFC1W25ISm\n5sjeZrD3NvdXHa/mNffn75CJZ1XX6DcDy6OGSbIiZBfUcs7Ypyw2N+j6gvGTpzi95ikHFGaC7SxN\nMFzNv05Qx93VCLn+f/DbDTeDYnPLvFHsjVew1RXj9oLfaB7SFyWzOqd+/C7nwyEnfsHhqTDOenJZ\ncR3GVDIiDKaMyxxpFtTuXUwIjJ/1bM4tl2+c0M+O+PHhT/CN9/9P+m1BCfhRyyoon0umZFcHPEtq\nmuJX+fTwhK9cd8y7p0j5iA+qQ87yD/li8inOuw8oxPNucY3xgcNwwtv+lK/X93jU5oyHHzKVirtn\nc7prS1sXDFWRdzta9bibU3x3k7HxOJNC11CcPaNMDUVZcipKI0pGQK6EwzKl2VxydpTyMAw4kBMO\nNi1VplyXFxRc8ah7lf1wRegdk8WShQrTvmHlMowIo9xDFxBG9KmDxLL2AxpR/MU5k9kh8zqh3Jwy\nsTX9foGKYVtUdPkEu9iyWTzFHQ8YXtaYqmMJ3EkMTnKuD8ZUgwm4PUZVS+g7uumE9dM9LsorPhhf\n8fbeT/DFjyBcwScgYCXJkFde+cvcv/9zbKp3GQ0/9bJLiqIo+lh698u/xPjwiBuvv/myS/moPQFe\n+b77d54/9pFTVUK1ZW1Svvytx9imQ0xB1kE5V/7hsy+zuvE+55c119ue5N1DHh68x9Dk1DU8e5bw\nue05w+yzjNuEZJuxTD1qDdvNp8j8NX7/EozncmgYm5Y3tmMaHVLYhGTWcHc+5KTIuT5/iLs3o05h\nFV4lDEZM3Hvk+zepltcswxo7v8ZgGSUpvi2ZPVkxPBDuO494S9M8IciE7EbCZl5z0AWc7VjYmtRs\nybNXqO7uMQhr2Hdk95c004pm8AEu8WzMmBubGbPudVY1lP6K7N4RWg1JyoIjO2LeNGxWD7gOCak/\n5UG1IhPLcNLQ7Z1xfTvn6OqU3yzvkvpL8o1hPZwwvPEZ+jSlBG6svk7d32T2zKLeUueeThJscUw/\nGTBsKvLUUtp9JFPmy32q9ZxBknBsMmrNuD/6HP0oxaunGqzZ2l9kYjz7Z4ZrzVlfP2B89CmaJKE9\nPuGDqw+Zje/y4eYd9teHsHjK1WBGhaXOblBMD3EOvv40Z5AlIIqGId9cdJh+RauG1dTzml+ySNb0\neoCzE4ZFz/7r95hUh7j35tT9AF8KFBvGo0CTXOHaHL2AziiSbEhTg+8NuAsOjGOcdRzIGWl2ynRk\neSB3aE8dIRP67TkmZBiTExpHWxoG1IRSKWVNYRydKXmWTthPhNfdNXXdswo5Z41hL8nYYKkGSkPJ\neT5mGzYcT+7wi/190nFJkBnT5RyMJ1s84ToxbI4rjueBvj0EPFNZMuMxgqdPehSPNzOelgXLQcmg\nLfiav8eb8lukoQdzj00HbvoQn27QW5+jtN9idHWEU0ttLaZ2vLV9H/IPudlZxjpjoz2vT3ve+1f/\nGn/yZoEax9/7x/85R4sJrz78FlcbxZIg315w/3jKw08Zbs0qEkYkPiXIKxwUS9ptoA33KOp3OG/G\nXE8fsw4lhTq65SETD7VPed+/S5fe4XS4QYuaqzZjr15z5GZ8MQT+sV3wqB5wVsx51Rv2Nx2n8nmW\nzkHT0OhTMn1EdvIqo/1DfuYn4f/+H7+KGy2YlPcYpgOSdsE6m/DoqmAmQ0Qh9IcUj1ZcvtpyvCjR\npuOwf4wNDV1acNJeodMBM96nXV1iTIYjJwsd7E9RDczTG2j+Kl1/TdWdcVzcYdhYFs2C/tmv45MT\nNNnnyM95p3S8cbjlQi6ZPTgm6Ydc5TPcWsjyIX59yoEuscEQ+p6BKdgMP4PzDXm1ZqgZzWpIOW+4\nHpY8eKMgH7iPbFv9Ix+wAE5u/Sz37/8ci8WvxoAVRVH0h3T6nW/xxhf/1I/U9Vc/oH8A/Nsi8veA\nnwKWL+v6q/add2jffY8vHb7N6Cu/zKAuudw/wLZrwibhkJSr6hCxDxl0Bef9KTrwjOZL7qxa9qpj\nkqRkKpaDFppcqH2D1YIkKRm2hvl6QE/PUANFMiRrpkzKMeM6Y/y0o2sMy8UW1WPM9px8CxN/zLqq\n8G6faTjn+pV75L/xaxjA2AEuJNReONcef+bIDy+4ctBXhu1yxeXKk+QFngZjAk/GI9CUIukp/ZKi\nuuREMs5Sy3i7YhDeorTCdfkECRkrv0ZtoA8Dsq89YPzKjzGWhIVrOHVXzFYVfpByHeYIFYN6n21e\nk4dLejPjSbHlZnifKgxZ6x5mO6fvZ7i6YxAaXukOODYT3iVDjHLNmIldMTLHNLbBmJTaCRvZsJcd\noOGam92SpE+ZpMK4sLyH4ejgDk+yc+48/SUyd83TZMzlIUwv1uC3PL1q0LdfZ/9+YBM2tHpGEe4y\nGOUoDW2oGW4rwmxKPT/nOjliWD+maSfY1NKcp1ThCXvApoCaCuf2qHWAS7dMwoIubamejPmOv+BO\nOSUxgcX+Hs3hLS78nND+L2zLgjR1iMxI64qjWUL3+Ih+XeF8w9XiN9DJJbntyK9H3Ll9xMNBRcgW\nGD/CAihUJz9NGn6NwnoWZgZ6xmGxYJVuGcmQo7xn//wbdOyxKA+weUb1oCOf5awyT1ZbXD1HCkvd\nfos6aUiSmtvmJltZ0oWeRAXvID3rsW6EEgjGInpNmtwi9I9pk5w2G4LsMR81SJGxLKZoqTxczZik\nN9lL3ubMXpCPr/mge4tSHOPsmGoWcMGyHh/w2vsfINynaIVHqWc8KgjZmIcnh5jtgm+vjvnGr/xd\n5k8eM9GKhRvQSkAaTxZ6Jlc1dfZNriaGgkv+TPoF7h+Med89xdcb8nbN+GpJGCu/1R8yN567/TGD\nRnk6f5/ATa6O5jwu7iM2oeWE90NJZhyDpMHlDW/6Aedmg+le4YntGZkC8Rt614MYrEK77MnknGJV\n8Z1kgJ8+4dniFLNcsx3+ONVQuDu+QTG7y/qdrzJyGRPdY96u+LFzZROecig1xntSPNQOpwZZrkna\nKzzgnKOzx0CNWkvW9NRdTe+f4tIFLhGCWFRhbA459x/QuwU3BiMeNuCOU35haNm/SNGupUunUG8R\nMXQIobzJeXOLNDRMxTC6dszTQ4Z9x1QSpFkx3V5Shc9w5I44WIy5G4qPbFv9iQhYRXFCkozZbL7z\nskuJoij6WNouF9TrFUd3X33ZpbxwIvI/AP8scCgij4H/CEgBVPXngX/Irov299h10/5vvpxKwZ2f\nowrf+NJXyELOT2FYrc5omkAvgaCCppb9xRGddVwlK67Pphw3T1nPT3jFW0JbMqxWbMoSXE/l50zN\nAZ4UqzPeqNc05ohnr+3z5umcaXrAhesIISFUniCetu8xIqRPc9qJI9n27Iuh1ZryKeTlAtcaLsKY\npAuY5BBDoBDoNXDZ5qRdRV43eGsZrTokTVEd4KUhSS3DfosZKPvVM6QxqB1QujVjM8BUOX7QcVDd\nIGjChd1iupapt/RmwAfrhM8dpdy/ekq56ulVGPSByvRkFgzKsDmg5RGZ3bDyQ16vB3w4XCHJhlWY\n4bueVFreuvoMt+wlC31KluZ0rWK6jiIELBnFMKNOa1ZacuA2LK96zGjE2jimiw4lQKq8Kp4H/ZLj\n9Tt8+uIb5Ch37pbcJ4P2MavkNrkskdMObRoK9YT9KU5uoBtHoilp+02cOUI8ZGmgDFeoFvT9isNk\nwoPaM7mY4wzYeo2ZpDwpjhmkCbaZcZBcMl+/T3fwgP32Ni1TvIWzbcNPLyv6dMqX/Bfx/X1+zN+G\nreNbkpPVb4J7yDY0bIEkKCI1vULbF5Qffpvj6YRtP8dmjuv+kFoE0zRkR1/Ab7+GhgGBm3yQPGIc\nOtblgleko+t7ztXx8PZNfD7iMrnHrPoGg/UacJCUpJunbHrHeu8Rb/HjtGZNvT+inJ9haPB5Strm\npJojqdCJkFhPIQnvD99iPsw53HhGpsOkCW06pqwT2tTx9XuOP/kkJ/iU0z3o3F1GfckiUYaywIiA\nBRc8wmJ3paGCqrAuJiQjQ68t+fIR7WAfzp5Bq/RJyplPcKFmzzRYXzJYt/jBmuvSca/PGMtDrD+B\n9gwzv0WYP+bKTLnZW/okoQyemiPG0jDqM84zeOA/wzfNKVmnjOuOe5crBvt/AkKJSKDQAa1aAvB1\nEd4MKUNdcT4ZYqQju87BeNrNBXvh25yennBWzWm3yulwxc1pgcOxbTv6XJjPhGwD1rTs0zO5bNiW\ncNeesApbmtyw6Tq2DiarDa3P8ElBCAFrEtqwT9JtWLlDVr5kaQNNecqJszg7QXzJsF+zocCgnG2e\ncD5estIJhNfZppYiV9rO4RRMaBjoCOunbA9uM754h1Yb+ixldPUUV7QsbUoRVkx1QqkLrM8o1nNu\nXb71kW2rPxHdtIsIw+FbVDFgRVEU/aFcPtoN/XFw50cvYKnqv66qt1Q1VdU7qvpfqurPPw9X6M5f\nU9U3VPXzqvrSBrdS72nD7tKuu6HkUW9Ilh7rHZvKUTWeO1czEs257Evm24TRcgHL3XU7lekZ9xWd\n7/BVw9Plt3DJnJU7ZdAmpI3l0H8O005pNyNm29u8Yo65Ye4x3ggShLp3uL6haxc8qB8zmo8Q35D7\nlsKO0U44eZSB9FgCvTukTmpUA40xSL/E9p6DzZiR3EDF0wfwnSKhRL2SBUuGI+/WCLseyNqqYuwD\nIz+hVIsKgOBbx9C3BAXvldDVLNdbvrMU+kKw2oE09K2SulNmOsaRkeJwBFr1HHQZWwJ721vYzJFo\nYGIT3m7P+fHQ8HRZUdVCbgRnhhykHYhlowNyN0GAS7vH/XBE5wPWK0pgOx2QTTvGScex9Ew3Sz57\n+mVGoSdoRxtqBmaNs5B2C0BJ2pTQOzp1OPHPexcElwjXtsH1gbQteFQ7nPaIGoL39NIzN49xNuCs\nwZUZqQ7oxLItW/ZNASgzfwffBlQ86/wxAUj7lhA8ttlQbPdwtqdzPet6w2j1GhdrRx8cnQYqUZbB\nU6+UurN8p73NE3eHbSVYa6knDTJuKAdDDsOUafsIx5hCHakJ1N0eAaExQ0qzpL4pfPPkNcZmxMi1\ntLJg0z9g3bcQHMa3qHp83pGGktr2jGkZug0QCFqTNi29FHg1jNhjMIFjCtZliZSe+YHlyaSgmnn6\nRJlke7te5YwhSUfIYJ+rqZBi6JIRGq4YVZds1xskdBit2duec35jRd4nJGLY6JgPwuukfcWxX3Br\n81vMT59h65ZEYO5GyPQuZnJCMJC4Xc+Ax1fXvFJ51mbAw9TT1h9yM3mf0MNQBoQgjLoBgypl1JdU\n4ljgaVPYGktvDDeWQ0IdeO3hnGHVUlYttgvYXunyimGfs6Rkeuqo+oYneY419/GupzOOxuSAsm47\nVmfnPHM53iU8soFWn/A0VJx2G1bVN6m0YTtIaYuMhN1ruOUgU2HQOVwxwVnLIm2x5Lhwi97PqCWn\n1owwKxglJySSAQFxay5kSpf25HTM0hUhPMIET9p55mGL7QTddpyYu6w2yo1eGO7dAiANltvtksS3\neHXUe0fUmQABoy0hOKRXbpnXKe2UobUYDQy3PVcPHvFR+UQELIDR8C021Xe+13VqFEVR9IP7bsA6\n/BE8gvVxom1H6wLBOYIqlVuy7K74DpeApzFbxAl7zSGFWEgMB6sNWbXr8a6xPTUepxYUFvtDqtzQ\nJZD5hlvJCBeUPZ0yMncYaUl1ccZo05BritlajHME8Rh1dK5h2a8Z1R7F48QDwv7ZllkNWd8RmNI1\nux74bPuQEM4olnMsgQQwtsWkOaKGAYbiaMibfsQddwN1npXkmAy6vmXolL0AucK4TilCStImiDeg\nUBulDYbXw31S/yvM+hGzcEFJRW3u09o1rUnIfErpDZ9rTigdEKALBaWp0KTnKA3calO+YO6w6QJ5\nyAl9Sai2vFl2iPGQKK03dEG4trdRL3Q+48CvmYY1WXbGXuoYlacEu8FY4WjxPjN9hVw+ixPFVS1l\nqFjv5Qy1Z1RdMQg5QcFguGwCSVPj6ZHgcMHgjfKwvAddha8alt0jgnzIe4NfhqDs9nKUmU/IuxLf\nO7Llln79jJu9pXQJJ2ul1ZY0eAAmzDi/uM/V5WNse41t5zyrzumCgFGcS1l0u+6uK6vMizWq8G57\njz3/lIE/Z0CHcwEN0NiaQSjB9lw1h4zDMVZTVgSSbcl1KyT+Cd82KzTruJoJ1veYfssNf0rRjQim\nZ6yBIxpyW+OTDUEVT8fAphgPQocLK6xCtq1JncWEhAN/SBi+QWcHTJqO4WKOhhWNTSjtiDv5EbOw\nj5WEgStYjUo2hTLOBwxT6JMWtQZHTujX4JRycUGwymFIWeoNvnPrFm2A1foGxeVNtK3p7v8CIc8R\n9Uz8kAED0uIAFzJC8CRhQtaNeON+w9fDDc7EE+qEO+tXeZ0J+/aAyfaSvPFUCBXvs9WMvso4sx1r\nU5C1DaUbkdqekYGRCdzYDim2ynADmjj2WOJVmNuAV4Gk4mqwZmMb9PmevwuOlpS+qUk3NVf7CV1u\n+Gb5Lj0d837ByeJXqXUF6ukTGLoEAYQUVRiEOcX1I2aLFkygtw6PcDmA+WRMW1i8zYCSsfckpqST\njqBQkiPS4lCcBvbYkgdH8A5p/a7r+uoKHwTrLb2f0w328dMJGww5WyQEVgTUZAQxBBJsVTNNj0E8\nghLUgCqph/ef/fJHtq3+xASswfANnFvh3OJllxJFUfSxc/3kEcVwxGA6e9mlfLIJ/OI773O2eMC6\nvqK8eofl4hq88NAsGPiETFMGknE7lBy5jKGZIkC+2aKdY514vIdgDWWnlOsa/3yUJfVbLjZr6uEe\nw95i+o5laWGwAAAgAElEQVS2XxHqOSEomW+oR1/jrHuHJ/4BIFxvTzEKwW8wKEdyQOlTRmvlDT7N\nTTvjxL6GoHShx9MgfUOtOapgvCfva/aTEUdasaczRq0gzpFqhhAwBg5sx9SMSVQQ37NppgT/Nr0Y\nBEuQDq8dznpqd0niGn7i/D4TJvSs2CtzMhmy7yALhlGYsdF9rDvBhI6EjoAyGY0QYyi3K/LuhFa3\neLfFdxVVqLmsLeodoJhmTd15QhOoOqENCVY9hdkwTAK5hT4YkuEzxmnFK/2EnA5lNxZSEhzSjxin\nR9xIP8tt+zZl5xn0CUNfclDNmOpmdyG/6WhCShE8xeIJBKVxDtPNGTVPsc6QtWsAkmDJgdTvxmy6\nCkNc2DB2NdCgPO/KXpU2PQXvaFxD17UEp0gNxWLA3HiCCazDkFpH2B4kgJeGD8MhOR4kYFR5p1jT\ne8AFgnRY2V2F0oriAhQMED+CLsDmCNPs0fkOgJvdfaSqGSx63lxCEA+akLc9eaL0SUUwz0+3FIfP\n5wzSDc3AI31NYnuSECibLbNlw/7FJYedYF1GT0HvFK89nd1ixPDW9B5DsRy7Ma/We0z0ghB6VKEw\nOWIsIbUYGiTfEPr3CNkF2SqwDSVJSGnyFDQw0ikahJO655XmPRZmxR5LbspuedYnjL0jLRxZ2oE6\n6D2I0jQW54Rhd8BEHCKBIxkCHSYoSZ/g+i0qKXc2OVkfOFj2uzGoNGC7lJLPUZBjfEYSwCjMVyfc\nfXyJd5ZN2rBOrmiN0mbP2A4T8B6Psg1AcFivWOmRoDzF06Rfh/CbJFoy0hIhIL5jkBkGrYDPSP0T\ngu9AemrJ2Ju3+OfX5/a5BZPQ5jlilAPtUITeCCqQGGWmR7gi4MQx8oKj3w03kQWsU249XlLNt3zq\nPlQrh+87NM3Z2hSxU9qgpP7pbiw/73fjmincsW+Qh4CGgPOQS0JKR9MuCdvHH8FGeucTE7CK/ASA\npjl9yZVEURR9/CzPz5jeuPlJ7ODij5Vvz7/N2fwJtt3gthXiPaHfQrVgSU3S9iTtluACpoNxUExf\nA0IiCaZ3pK4m4HFiUN9iQqALDc6E3SCoYctFCbdWG9q+wnmP45LKeYxWONMiujvy4W3CVmpa3dJY\nj1gY9Z4DV6Oa00qOJjAOQ5aThPk0Yx5G2P4WeZcy6CGwGyD3IDEM7E1GdQuhw2NQ8aBw0iUESUnc\niOAE0YQgPaFPAUNKRmI2uBA4LVquU08XGjyBTgwHxWuMzT4FY2a9wQVYdiPg+fOrkEnHlBUkYTem\nkW+5etrSbbe7Iydhd3Rma2CWdKDKhDVZe0nT7gae9QJeUlJZczuFw7xjO52B1PShxbotHk9gN+hv\n7gfs9W8zkN3pT53kOLcbkDnzhlQ7JjLbfS6iWPVsg6LNOXa7ZG46WrkEGkb1hpE2pOoogwMMRtvv\nrTtdamhQUsnJdfdFiRdB1SGqVC1U7S4U7G0OmJkp97afxolS94p27PYaFSRYbLdmenWOrGu+Nrxk\nLhu+nlmMCruZFYLDicNrApoz4wRxDnwA76m1xAiMNKFwG6Zu+HzHVNEgGIGu7fDpdnfhE7sxdrtt\nRxoSauuwPmD8gCxkqFfUe+ospXcdeOiT8S7cS0OTNoTuGXN3TbA9ScgouxLvU7LQ4L0yciXHcsBU\nCoZ2RNoXWByIEownMWtq2QWDWXNOUEPvPXmT81q64k7f0ZoUVQc4jkPL9vaIn9Q5WecpQkfjFRXF\nqGW2GVBuB+TSMXIbxlgIa2qz++IkbTYkm4eQHrNvprvBsQOYkCL6OjM7JcHRh5QuLLAEPt2mZIxI\ngyKu5WxwhUogkS29GWLYjTs3snsYNQgBJ8pacp6ZgnS9odALhhyzn+8+y0GdkpjFruv/9hwbVpRd\nz7AKyG74bgIw3yuozIzGZoiHM5tyridYPIpBRchdR0+KCiiBjIKytVhJSIJD8fShJ2kdB+0B0jaw\n3uIJFJqxHU3RwRTaa16Vh6TFkGJYMrLD5+M5Cx5DQOjSgKqSYrH+ozsD45MTsIrdxisGrCiKoj+4\n5cU506MbL7uM35eIlCLy9suu44eh3qxZbVa0gwHHyzXT7SVehBAclaYksutljrDbaS78kKABXE8h\nCQkJSb9E8SyHSmUd3z2hrKy2tPkIwWINqO0I7RYVKKQm+HfIXI3gSLsNJnhcArlusUngKuwGZ0XB\n+ECSlCTS7epTcOQcNDlVXWJcoCCnbAODzlOaAbfUkknLYb9CpMUogGJDzatVz8KlnPZHgJJKIGCx\n/pqu60AhQWmkoMo2ux7GJCerKr7bKXNjHGepss9NjM8o+4xFCJhQkfoUBYbqCboLpLNugd/0WDyE\nABqA3dEgO6gY8QDjAlYDSecQ12J9S7bJCE1NEEveHVGRsMkG1DaArACHIaAoXnPE5budPw3Mjwq2\nA7sLi3jSYEgk7I52iQVV2nRDQJlev0++WoLbHQFyBI7XA15fnWPDNeb55xqe92VW0lCkGRd2SO0z\n1AnslgwGUm+gVzofCMFzy91DgSCBVlM8Nb1pUMCqkveeJmtxJmFbL7nIWp5luwAKELTZBTEVEs0h\n5DRBoHO0kiGhZ1j3HF7XPK5y/vTmGaRbguzWodfMIYlpCRLY0CEqeFVEFZdYjBNyGYCv2AwLcIFR\nF3Ca0BS7oyZJ6AjOf+//x0lL065w9RUPnj0C2aVFAfAjEgkED0XlsMFTBEuqOQNOCALOOnKB5c1j\nvnN3yu18wxEbFMGoweHZC1cM1xtwjkNucaOHMljIU5rxgLthhSWgahi5M/a1Z29p0HnNcFGTdT0m\nQKY5VgtAKJygxiA4Bma3Tn7m4oKiH+EtjGyB0YB6h8USrLLNc0QCedMQfE9aJSROGeFwKiCCtWPM\n5ISj7G1QWJmMQCDxCVsZYRHulyuMOvrkAjSjMDmle8LYb8j7hBzDdFHyqfQeh8khBoOalH56F9FA\nbxxlyHbbr2TGPTF82t7iTe5CkjxftxVfpkzT29zK7zJwglXDoDO8+uAc62G/Lhg2YL3HaqBLDYkt\nOTIHDDUwH2b40JPmu0HEVdJdyM4SehMQUfqwpHv93ovcJP+ePjEBKy+eH8FqX0rPulEURR9bqsr6\n4pzJ8R/vgCUifw74DeAfPb//J0TkH7zcql6MEJT/90u/xPXjM5wY7hQn9EkDRlBSSBxb1/HUdagq\ntu9JNWPIlgEOowlJaPEotfaoMfTG0zuHCw7ra0pdYXEkzYK982dIcKgGJjxmoXNEW0SFsdsiImje\n0YwtzdQQJhZJA7t9N4M1GROtGLfXJCElYHiih8zzASrgkp4maRgQGNgp5f4BwRiCMZQB8mDIvaV0\n93Buypk7oNFyFxuCYtRhgzBvvvl9OzKCSdbY/j7b8gEqUOvuVLh17mkH0JbCEEupQm7g3fAZvh3e\n4nH/BXCvI/2Yex9eIK4noaJkg9HdN+9Wd6fCSdJiui1l3VK4jlUF6caTb7dMm2sWZsBZf4cLjlj1\nBT549uoRRTelDIGh20Ub1MLWsbQlpStQX2L9BCcBlV1Scc+DkgADr9yxb7PuAi4ERusVWddR4gCP\nUUPpM5Qeb4RgDONg8AqigYkd4hE6Be9BWqVn16Oi4nHqMKnFYUm6PdIuwaAYCfTtU7LasfUZ5K/i\n+1tIgFZq6tRBABtAeotKSjApzqyAhkQNqiDeIz7Qa7K7PkuVZm1gXWL6W0zqAwIBIVCSscgvWGZr\nbiVjJm7KNk+oBgneWmqrqEJd9jRZinVgVUi9I+0qWnVo71HXYpoagiLOg4MOx2LT4DTQyy6AGXbx\neeyS761LPJ9igsWQoBrIVNiTzzPRIZJbQioc0DAG6qpj06fs2Q6ehygbdmeuHfk7OAHL/8fem8Va\ntmXpWd8Yc65md6ePiBu3y7w3M4sy2K4EyqaREKgkA5JtgRAWEkggQOIBYckynQQGYd54KgkDQjQP\nyJKRRScknjBySWCXZVMljE0VZepmVVbmbaM/zW7WWnOOwcOce5+orCbvrbxxb2b5/KGtc2LH3mvN\nNdfcO8Y//zH+kVhYIJpw6s+Y7c4wy4zjgFvdDpCyZmMysizopCWiJaUw3tDYJwAcD2dF1MuRMGU0\nZ3zrrGwJIvRxSfAda23KlYzKm/nHeCM8oOsechxfo90mwlQIv6A0ZKIpY+4wLzQ97JywSygRPCNu\nnDQr9hQ9JlAz5rqCshLJ2cixZdSJSEN251nYoX1L281YNS3LbkIRHLjqV8T2AhNj6mfMhkDcOCE7\n0cvNWcQHXDz+hDM7AsBRGmm4iKfE8zVmiUG8bF8IjG1gbCIjE65bksLJ0fwH/Sr+1PjbhmC1zRmq\n7Z2CdYc73OEOnxHrF89J08jRvftf9lC+H/4D4A8CLwDc/a8D73yZA/q88O2na95/vuHy+ZbrtWPb\naxoS4hOKswQuhgVtaFECAgQCMRlWCUlrGUTYbZ4yTmuG6QUpD7hCHC/pfYuSaFxohkwgIQrmxvnV\nU8RGQtzytcsblvKrPNSPyAE8JJrFrxL0Mbgwupc+OPF1RCF6ye7KBHI6ZhXvoVaUis5L9VcnmXah\nxDN4S9e8pRPv+Mgb63t8qG9gQMucyYRTM4KUGqgLeUyiYaIneMNIQyNXIAMuwoSAGzmWnXJT52lY\nYYysvGXuLQ9szv1hjmanzQ8J2fAx0UqLOkQmoiv4gpnP+epmSUxH5Dyi40i0kde4z9f9NSIJN5g8\nIBiNZWQzgvWoDwR6RrvPlhUDkU/SOUkC03SB5iXUgFMo6lHMRQXqGFklK0F+UiwrYkY/TahpfY9x\n0ynRRkbdkYOzPTqj9RUSWnYacHMisPMWUkmPnPiQST8htBskBsKwI2SnyXNwR6apKIoWECJvhtf5\n6vzvpGNEdUesqlNc7/C8hWwc2UMaD/QMTGKs4wQ4uPAwL3lLXuOYI8ZsbEk8G99ibo5nUIxGJvr4\nmF3znCzKyk+hTSCBKADOVnYMDt7AHFAXVITRJ7QaG4gFPBmnV4HljZG7iGNkUYxUeZQUIc8NF2f/\nRxzMyzWvOCbgXPAGTs9D2fLW9k0mzagaoARb0qUVR/YuMxY0yco6N2josIViTaElDRlxx4dMtsxo\nE5NtEd+iQJMjjjDoDDVD6mdokR21R4Cx2mXmU0vjgeWoLDcj2+T0uQF3FvEckYCGnrf1x3jb3iLS\ncawdrN5kGx1EyprKwnxSMDiargkuXHqxTRdzNt7jDu9P79CScQQLMHmkT4k2CepjsbSnkItte8S0\nWKHaMHnmRR9IwVGWtBoo6mH5rpg88PxoxXoGV6vAuj3iZH3GcjgiUDYCEoJrpLc9CS7K1M6NY7vG\nsxN9QiQx4YyhYySTA+AbGpTl7voVf1Pf4m8bgiWidN1rDLs7BesOd7jDHT4Lrh7XHdMf/hTByd0v\nv+e53xXWsbm6w42Tc/ziEWaJblTe330HNefei+e8/eKyBD9i9NMcdWe2icQsPNxkTieYJUiAjx9h\n02M8CsfWcp5KrU7EMFXGpjlMnLkwssWnD+mvP6ZdJ96wyLEJof8Q6z/iqU6s82MeTE9BlKTOvH2X\nqYk0aiWGzQ0iSk9HtBKIbRWy79jpRDftWPq6uAlG4YyMeialiX0NmQOrKZE94IwMoT0EMhd6UfqA\nocxsxjK8XmqpXCrpFExhK5FxcNbpGrOMTiNdHhEDTU5vzlZbztI5kksQ10pDNGO1zfjTR/R2RuB9\n2vCEIImlz+jSikiL+AzLTxhGmJkiU8It11TJOWs5x/IbqAWyC5YFTceEIaDZkQy9JHqZOKZlRcPZ\nUNSz5IGkG3Bn5iMLu8HM8CyHnXsXUM+oF0Jpy4YxNiSkGgE42YRkymycOOEBTeqJ3TGELWEoSqXX\nOeu228JtaJD6p3RZMlQSz+eZgQC7FnMheUuMxZzEArQecCCLE7OzsBk6zRCER03gk3bkV/cErEIx\n0JqUacCYOR1aOg/MZaR1ZxOe8u5mYLZc8+PPfhlHuZi/zXn/OifbifPdDafbCRNBkyIWkBjITMxH\n2Mgcp6GzQHBFc2YMl8UlAsFkQHPH8XDMKnUYwlmO9LojWKIbG47iPUCY6FHvWF87uwAzFofrCWUR\nEkKE11tyWZIcPzEkOTmPjF5c+YSROWUNG4Ij3Dx/DwARWA4rWlmi7rz5TGmSEMZMdLjZfkCTn9Ov\nt2AZUCYJxHZBoy09pebwJLeYKtdziC4oQjM5Z8MJX7GvFuKHc0l/uCWZUGinrVFtUAIoiDqr0HM8\nXSN+RZjeB34VkwlEWOQ5q52TxBlnK3bLFVK6DhQ1uMpgeQIdErtWmYvyXHqe+pw2F8Us1PTc2bhl\nNEH27zWDyenzGs2JxhL71Ne5XNNKPmw+eIwcj19co+FXRrBEpBeRvyYi/7eI/IKI/On6/Dsi8ldF\n5D0R+fMi0r6qMXwvuu7hXYrgHe5whzt8Rlw9fgTA0cW9L3kk3xe/ICL/LBBE5Bsi8meAn/2yB/V5\nIJnDsKNh4A37Nea2Q/OEWSKroA7BnDPfsPGJI4HTtOF0aLnYCFkiEaHPAZUWPJeaFjE6u2bOwHww\ncEUDnMR7hOrnPHjLYA1TFAa0pFNZBkl4NbuIZhgN21Vbgnw3clQ8dogURc1zyY2L2YDiJnbVTmQf\nMc2cJiNOhnhTUrqAXgZMqC1WnOtpyW5acG3b4lIX2qJWUMKqjg6XyFfz2wyhBO0jxXgheNn5NgKX\nMpI8Y3kDXhLxZBRyCLilYqDhEIZC7gKJZrtglYxxgjQ6UQeCb3EvtTzRhDfDA07lPrtxS54SiZ5H\nzX2SrMl5xqTKOh7zIi6Z2QnqEc8dLsJiB31SlrtIkoS4MGdOSz7UpCFw0TxECSxtx6A7TuMDTuUh\nhhQyqKV+RnwijSNz1szSyOBT9Q4EVDFXnt78Ld64eZ2okcZbkt7wlfBWCaY1IOZECZyuDUHpCDQm\nNEGIJqRQzDRA0dSQRQnitJrIQavyGGk8Mp8CY1zTSUuTG8RB0wXBSgoh5mi+JVnPz4zNiZSaIZzz\nfMJb2ZgzFhJZCbMZTIuWTqpBhgiNBeK4oddNCc2royETnKZTznwBokTZIuNInIraFJJRsgYdMRCE\no+GIJp3zYJoXwuxGn1pknQgYjRd78JQHpuGMUZxTOa9X4UQyITkyvcsDv8aD07Zf4+i6YYjlsyvZ\ninEFQktCJbNnNynv6rGEMW2JY6DfFTV69Euylzo8d+fROON9vUewjPgW6Jj1pS4paflMze0pY7hC\nRVnuIBqcd+9wL77O8TQnuxYSbS2+1dpY2bG8q9bqzSG1z8SZ9aU2sjMIPKWdNqgmggcudMvQvmAd\nd+AZhOoG6axzy/EYeG7PEXfUDRyO82ntcec0DnNaltPE8ZS4WI+4CSQHMTpzNLdkT5wNaxoHInjT\n4oBKJkXlRd9ibc+VrT/zd+/vFK9SwRqAn3L3nwC+CfzjIvL3A/8R8NPu/nXgOfAvv8Ix/Dp03X3G\n4fEXdbo73OEOd/hdgetnTwFY/fATrD8O/F2U/3/+W+AK+BNf6og+J/x/n1zTffs9TvUSTcZJ2LHV\nLUogsMByh4sz39xwfvMhp+NIv9uwGAXhsugO047GqrFBaAGllURDJjQdN1yRQmB3dEyvgUBEzLFx\nQ9ZIoxMbDzzKkQ2RRCSJomTUT8l+wpAjGjsyjrnQJGW/i7qzJ9xkp0tlN1pc+PD4hNeajp5PeE2c\n6HOCL2ugLSx8AHfMjATMbcfTaU4MEUHwNKNJkZUdIV2LiPKkOaVlgnKlpQcRgAsmgrnzVIwtgqbi\nMBYscu1zxIA8sbATMCs75UAjA5YCOQXirhxbCKzpkBJeo2JVe2hqrQps44LrZsXjpiVOkIHROzJK\n17zDrDtmOR4jHoFANKXVxM53qCpLZigJ0at6PaWo56J/yLYpxgXqQpdLDZK4MA8rjptzXEaUVOqo\nMNSFWGuLmlQC6+txzuW2PBeZc+GlXt0KoyMmZ7CmBtyOxXmZVUn8He03WaUeMCTUKh4XLvWKhkxP\nUSPdHaHBpKTniTgRZ9gJ194y0jJYMar4OD1GPeBtxzvpHZZ+RA7VeY6GB9MZ7k5GuWkXZC1dmQbt\nWYYZ1PW9yVvUDPUtcdzC7qaslynzdnoD8YROA1LrnnTKUFVVsYwYOJnz6xG34nZ45LEYY7iz9IB7\nMSxpcwml23zJvtWqVsWl/J6KccVgRHcW868S9ISL9isYT8uYXcAHHKdRZVZ0LkJtKh4xrmdzTIRV\napibMcSyEZF9wovgR/Irnl9/iKvg+Slni7eIHhmq2YkLdBGcicYKsbMmQH8EFIIzv3ayGG+sL1gl\npTFhlmaY7RCMrjkurxUQacjqJM/Mc6ni6oaMhAEWz9EQya1y2a5J5DqGch4HrtNj8IwLNGPZgOmm\ngHtGkrDxDqkpz0N6zA2RMEyEbIxutMCUDRsyU73GcdbTNJkh5sM1zwfjxfSIdjX7DN+6PxheGcEq\nje/9pv61qQ8Hfgr47+vz/w3wT76qMXwvuvY+w/jortnwHe5whzt8Btw8e0rT9bSzL65A+HcCd9+4\n+7/r7n/A3X+y/r77/u/8UYEjyehuRm7oi9m3G00KJUhB6HZOGG+Iw0RDBIc+LlFVcg1AQhqLKjQz\nAjAjMbZFXRGgpa3JYGCeuR6+A9Q0G+BFjmSLrFmQGAkkRis9gVIq9SMq4C5kQjHO8MTXU8O9fJu0\nMtiOXWzJC6VnLGeQ/ZUW84Vl3hG9BFBneWI5Za7GxyQvKZMlxck4zjOObIYA29CV8Rp4rU96aqdc\nc8aNz3FpmDyVChAHvDRILVfcMYY5p/mIo0mZpBhAzDyjUkImcSdgyDYzWKz2AKWnlWKQtkRy6R1E\nSYV6LivGGLnqZuSqI+XYgAnBYU6i5ojRSOJ8NtDFBYLThY8IjChSA2VHRBERZLqhnzKLYaKxhotx\nSZ+d6BuwibfoeNsFVYieibVGJjgM6+ecTCeMufTl2oURzQPgNUvO630QlJ5gkTaeFsXFoZUZX2m/\nWWqd1LkvD3g3fI05kejF0EH2OWauuGda4CjPMaymsUE7gWyGQlOtKieN8ub0Fg/SA7ZtZGyVnIXG\nu+KmGFuuZiueLE7IEmlDxLoZtD0tA9s88gjnaprwcVutyAX1QJuKAqLZ2MmeLMshHc5kCwxEU7pU\n0uWiK7PW2FqxBZlZ6S03MRI1IDiTLZCwKBTbb1MeV3XJz21bnpMzYq1HK14mxeJ86c9pPIEVRTCY\n0qRqUAEQZ4yxHCymzMl4TXZhAqa8wwCVLehAlggS2PYNk+bDZx8UzxHxYkgTXZgaZewO2iYxlXrF\nIC1hnNEORzTWYU3gDV8TqtqIgNX5mGfhiMTMMjPLNFVxdnFintUlUFL7plCyuGWCYA2tK5ftCdtu\nTpeUJzi7tifFFU8WDxG5AZzExMgJPT1HWVnmhBGZdjPyR0UVFxeG4xlP2vd5r/8lLAqL3HASH3C8\n3tCm3yV9sEQkiMhfBx4BfwH4FvDCS3MAgPeBN36L9/4rIvJzIvJzjx9/PqpT293DbEfON9//xXe4\nwx3ucAegEKzl2fkPfQ8sEfkZEfmL3/v4ssf1ecDd2TYB3U5smteQcAoYQWckUtm4rylkx+uR690T\nBkpd1TweoZ6AWAv3S7Azk0sWcaq259CljOI0KdOH8t6tOimWQHHIj0taEIK4kjcdJ0NRVLLDwJYr\nv6bLa5IFJNUyds+oGUdE3q5OY4LQaos4BDO+umvqwKCyHoSOq2nBW/kJJ/ma8zwSpr6OocyJlEtG\nzNjFgejQjBmLQvTE49kN625DNEVp2NJjwLBPq3qJNOmBGCy5mIy5lesXCWzI3AtV4cjFhu86X1Bn\nnYZEQ0kZCynTSiL7BsnGYjC+q2+w0bJBsZMW0ZJqlqfXahxeFQ8RTANtXNMCCx/ph0Tjid63FD1N\nuNaMde9wnGf0yapKVQiRmJKiVjoMRw4TWlSDlwL/bd5yOsy5tLHaDQiNlZ/BbsPDC19QjhCYU67B\nvM6FKL0seJjeoNd5mQuJUE1IvI5Y3WhQIoI4DBIYK6FZ5TPO/AGeGjSBdXUjx4WGnpj2TWRvx+S1\njmfUyCAtOW9rwF/u68nsbRazd5nrHPFA8AYQftzewnD6YaLJESMVVc2NJhdSM8bIIm14nXu0/Xmx\naBfjQ15H9E0sO2LCzrZAxqyk9wVW9TMYqklJGeveMn+DcGMJYpmPQqJy+RzbiOC0GKMWWjrurnk+\nlJS2SFH9ggm9LMGdlUtdt4KKklHmuiGmkSANKc7YxDUWMpGGSCZpwJc9SsKCMTushu/5vpGyiWCm\nWC73cXhp82NvDmICVq9QHGY442nH0lbkKlMt/YhMQN0OH3EopW5HwzXbdoWrsG17NAVclB/Xd5hd\n/Dg3scPytjRFHhKnzUOEQOOVtNkC0ZZsxbBDBVClzWWBRBpmsiBI4MeOTvnau81vcrWvBp+KYInI\n7/udHNzds7t/E3iT4uz045/hvf9F3YH8yXv3Pp+0lK4tDljDXZrgHe5whzt8alw/e8Ly7Pz7v/DL\nx78B/Jv18e9RLNt/7ksd0ecESxPj0w+QMXG5OmOcLYgidHpCqEXdAMubNX3ecZx/hTlXBMkgykwu\ncdsyTJeMTUkoiboqG9FVOYk1IjhvLlDdEciMBLa1ldA6f8zT7S8CQvaIIcxSKCl2nnGcx/ZxVWKU\n7EqN61ErKUVjuyOH6iDmyiJnxBIPUg9SAk8RmLzlJh9jdLyen/JG/5RdF5n0ghQMCz0ZJ4siEqsR\n3G2o6AgxZEC4zC/4gGflHwR2WnQZCEVBSrUXEgk80+qypJOVl9PKjsknqOl54Ay0zHSqNvkloJ7Z\nWOayonEnZkNTpk0wTdeYG4axtg4QzHtCbbEltS4N2RLDCzqZaPfEGSA4JkISITGyaI45796o1Egw\n6dlUIWblAfXIKgVq51WmNmLs+0KVZseDBmK4QXCi76+YQxQtCFvbEu2Snh2xWKQglRDugvFGeIfz\n8Rf8wCAAACAASURBVJTj2rxYDMScESXve2elEQ9lfsSdQQJbCRyFI+7x+m16HsqVjIjHmm43P4yn\nyYEgVRU0Rwk0KF0u6XTn9hHRXpBtwIEsyunqK9BclDoiHMmOCTRklrKrKqPjGE0ScmgwDZj2tLRE\na0pvZYFEU648TagLiQFJyiYlBoTLnHnaZDwY3g80IeCkUh/pwiCRX/Y3GYmEbIgHNEViDkX5rK8D\neLH9LildcmM7xBWTSCDTxCXL2dsEEW6kbIJMKriV++GeGLlCPNNaW6wy6o6EAssU2JpjkoheyN1+\nxR40rL2SZwZW5sY8M3lizDsCAT1kQAaG8RlZjVzrm7Z9i0/bqrCXk7sLX7fX2FnkRSxGE+0UwTJB\nrKZ6QuPKarOi8YYVO173gcXNyIhzv32HuS5xE3YeSyPhOo4gpbdXMc1wvKZWmoeysaALdrxOkNvU\nzVeNT6tg/WfVsOJfFZHjz3oSd38B/AzwDwAnIvtqQ94EPvisx/udom3LbtM4PvqiTnmHO9zhDj/y\n2CtYP+xw959/6fGX3f1PAv/Ilz2uzwM33/mrpKffYiBy3N1nF502LJilQPLShDaOiZAzk49kdhw3\nj+jDmmxrXCZyXnMzfAyUMDrHiNFyKSeow5r97m4JTlxgmwY072umStrXTnquWTBqYCcr3ALBi+20\ncWvTnNgAsLEtrV9hvldRBFDWfVMCIFVcjm7pkRtXnDLmOYshkAlMKmxCR6oueQ0tqSYtGk7ZsHaC\nRx5wQfKAakbE+UBesNFrQm5ZDmfVIQ9mY48fwqBMkj3R6nCEjRhJjMCIaUdrAyCkJjBaxLLQ2oxG\ne4I7ak6bEntZbWalH1Dc7ci7Dc9Cx3ObkcUYaXk8ndMw48haQm5BSl2ZyoRhhOzYJKR9/QyCh33x\n/n7cpR9UpselZ9eW43Q+53iKpd7JA2/6x5zHHZtQ1JCWhIlhapD36plxU6MzEdjaGhySjQQzem7I\n8bsgW1QuEU+lMbDoQYADSDonTJlAxiRhAuKGyViCfUCtKJAP9d4hsM9EhMCOjFokokwEVJQutczG\nQO8OBlkVEdi0mTGmmmpXSUYlkRnDfEAlsNQ1fUiHpEVnS2sDYVxXRdQZbEdSJ1oh6zEH5KCaSUnr\nq2cwm1h6j1ZHO8dJZIZQ3O2sEUKMxLatZhSRrBH1RI4tOmaCwdHWK43NleiVFWgvNUEOLtyIEVMk\nNS1JyxW00zXiTucT7kbAGWLEyTzNLwAIh9TIMvLBI9cOEMu9A1JdS6G+5mSc8+DxEaMY4DQ5IUAi\nk+KSWXtGALIX5XMblSxSiJgZT+0cXnzAcigGGUNNCTSEq/4FqRL+0+2MD8Z3eBJOymYMgGSiBKDY\nvl/FRdm0cJjrESLCNR2XVYnOYoy25nV/HT16jakRRskkO3yNoVkYQo9Ji+bvNZl9dfhUBMvd/yHg\nnwPeAn5eRP6ciPyh3+49InJPRE7q7zPgDwH/L4Vo/dP1Zf8C8D//Dsf+mdF2ewXrjmDd4Q53uMOn\ngZuxfv7sR4JgicjZS48LEfnHgM+8KfjDiBfPvovNVmxnxWrZxcjcJwssfcO+arxhInvAzNimOZaL\ntbeIYw7Z4+1Bo3Azu+C6e4NROiAT9LYIPHtgM1ySXXErykPLRFo0PD09xVVZHWWa2T2aff2WBa5D\nxkMxA3CRoki8HJwSMJSrvgec+eYBgx3xLF2gWCFdOIsB+rDARMkSuLRZUXjqWJz2cFQBXCYecJ/7\ndkHKCzpTmpRJIhzbNfdToHFl9C1gfDA+pR9KWJtND8FtdiGZcRWMrQpqJUUtxEyTn4KVeprJI6vw\nLvfCw4MCUeq4vBI3qaF+JNtElpLe99hLnZOLEkWZzMipuqxRnNmYJmYj7G3VofRSKjNIfV1p6oo7\nLbNS3xIbrF8gtX6qiytOfUXnpV7F24b78YSZt6VO7qUjiinuRQUIOOalT1Q35ZL8qCsmFdArvn41\nxxt4pk9xnHmKLHJgoqGTYlG+5hIxZds+Zmqe42hNM74pJDnPi7qTzgq5lMBiWjBPHctpRjcYISfE\nneNNz/kQuJeMm3bOTTcv6tqirtdKhl9GkrLashvDosdiLGtyr/lmO5hclGMIs1y2CCzU14ijosXJ\nUQyRQtotG3OKscyzfs3zdk1iKqmrAkpTFFVxFGVijlNSCRMNsfaeitWIQyilokq55wsizbhDqkQz\nDp/gWtSq/QIQYJZHqvE7xze/VNYF0On3GHS7sPOGnRT7l6wwxokkelhPe7yb32a1rXTSHE2ZKH0h\n+PW8cwzESTQIyhQb3JWbPpODI3niLC/pk+O1mfMYdow1d/LBdkX0hibeR/2cuSxptcdECWGF16RX\n1TnXRw+I8zMaF9xSSTH16fB5bXRO7y1te8TVquFj/YCH+Q2O7R6DPSW4c2nrshnTfnFp7p+6Bsvd\nfxn4U8C/DfzDwH8sIr8kIv/Ub/GWh8DPiMjfAP5P4C+4+/9S3/8nReQ94Bz4r3+QC/gs2KcIjuOT\nL+qUd7jDHe7wI43N1SWWM8vTQrD+nw8uS777Dyd+npIS+PPAXwH+db5Ap9pXhQ+3I/lbf4uFT8zU\nQIQgGZXITjLRXwoSHQzDZWB0xYml+arFg1CxR4yluL4l12M+eukgVtPADEfob0ZCDeQadoyeQEpB\n+3VckWdtCfzSBU8DeOzIXcPz/AxpZgfdoFiPld/2BgNE4yM5wRAu0znXwwnz8faauqkEnykbo3Zk\nj7QTpcYDyAdDCUebQiAVUHEs75P/nMYFNSN5UdYGT8VsAUFecjabsjJZwt1JWZEciEAIU0l3y5me\nXU2bikWZodTaqIPmUpS/D+WMKrsBV75jkqL0mZR6E3dHXWmr32ITMt12PDRjxp1Neh9BS7NmFPdI\nwpmKJz43fcRiV6QnL+SoULyA5kDslgiGKbSxI0jHPj3M67qIBLIoKs7kkbVdA85pc87IjBwXiJaw\nMXoJzE/kAhCOxrYkKsrtfe7Wa7ocy3mrTmPAJIExhP2UvES+QU1ZDQsaK6+PZsTkHOeWLMpEx7pZ\nlJ5fqgxaSFNsTjA3BmZ1zsvVJxdwQ6w8o1bSRstrcvF4f+n8XSppcwcmUQq7ONVz8FSOyb59QZm4\nFMoxLnUiVWJ7Q8/76SFbOwEP5SdOs4Vml5mk9LlKwHrzHmO64nrzrXq/QyHLKP20xcRpQs9ExrUQ\nv0VckVUYvbgzpkYZZ6HOJ7R5JIkS05zoQsDJzDCRoizG/Stv578pwiudt8QsrGXGRMtcJlBlTunV\npdWmHZzsAREhx8iT/oL3jt5GqwGNZi0pjy6YGFnrvHrL8/kJN+190AUr7wmVSD7vT9mFDnBMIqfa\not0pp4uv0xjMQks/OdfTNYtdy3Jy5tlIEjGfau80Rz3yje2bsFvzdPwQ2IHDtPni2v9+2hqs3y8i\nP01RoH4K+KPu/nvq7z/9m73H3f+Gu//d7v773f33uvt/WJ//FXf/g+7+dXf/Y+4+fE7X8n0R4xGq\nLcNdiuAd7nCHO3wq3Owt2s/OeXS945/4T/8y/8lffO9LHtVvDnd/x93frT+/4e7/qLv/pS97XD8o\nPrrc8sYnHxCxuvdeVI4mJrI6M+kwFTB4vixB+iUj691l2QGv8WOsAePi5pq4pzxSVKYoZX86MBLZ\noaMRpYTFg5fd7BwCRmBDT3RDfWLnodQEaYNlI24G1IwpFKVg8vzr0scW6xVSU5LKbv3ERje3ZUYm\nnA4JTYEhBL59ekpMcCJv0nkgdyWwK/VCygTcVCIA0Eyl8L5NkeiRXKP4jTaMclPPKnUeBZcZQkPv\nmWglnWgfbm8S7GzgppoW7JMfgxizMHHePCRJQ5JiJb+312ttR8+CyBzbN4x1DrVQYyzmCjElWk9M\npgixFukHgoLvlOyh8FHPXLSJIX9wqwOKMGhRBtRhFBhmc1K4dYNTAkmLE6Hl9S2RMsPjbR1NW8ns\nnjikcMXabpi82GYHi2g85STe54jjQ81XGQhEaWlsT7wirRQS06ZIqmtIdJ/+JlzN5mQNvBkvUNmP\n96X6OTOcwDy1NDnThxmNzmi1h9ovydmrmOW9//s3/j4+ODom1fqej1crPjw6ZiNdqXlKxklaMfM1\nm+6GJImYDGGqK0HrarLDnd6PSD3RImgAceHKX1Ar5nAMj/DMnqD2shpUauye2AmXw2tMKJOH0mfL\nDVNBsjHQkNPAZverZFsjVowszCciTj/syj1cXhzmaWw6dPkAbY+YmoYpNIxty6N7rxGTsdzsaJkK\ngdUZMytzrBLYSlOUNBVolOvTfCBYyylwtGsPV9+mQoxplNfY8bpo6T0ngkqsmxpCF4/KXLQdja54\naG+COpOWDZSgGUXYhS0AL2TLNrZcr4rhzWNZ1yNBkr1/afHmXHcNXXMK2hw+P+KQTejkhKBzcGGK\nK67CjmRaCT3McqSTOSZ7u3Zhk784ghW//0sA+DPAfwX8O+6+3T/p7h+KyJ96JSN7BRAR2vbeXS+s\nO9zhDnf4lLh5XgjW8uyc/+HnPyCb80d+4uGXPKpfj98mkwIAd/8fv6ixvAq8/wvPmA0XhMUZm2qj\nbQjrBTzqjlhdL8k+sGknwrRmSSTKhNhNVTTKcdquwXZgJsQuspOWQVpmFOc1QVEZDwEuCgFl8IZM\noGGsOlGxJbfgjK4HpWYwZ+YgKUGrqEQma2nNaaaRNkzMgZlnBpQP7SknzBniVFLTcBKRjJFd+e7p\nMZNEvv32N/k9N1NV57SoRMahKt8wntslUDpS5Qjb2DHPp0x1D9dUIGecWBQ/K2mI6jWNr9aGrY3S\npLdeU7KJ5Dvupa4YLGiD6K26tp7N2WEsq8rW2ohIoqc46j2ZLfnW+CucMqsKmZOD8mi65J55UdlI\nhXRWNQpgkg7XSCSjeUdDxsV4b/qIn5BzXArJCHv+rCW9z14ir4KUptCAWMK11D2JZVQEzUbUckfn\nued5KCrPDkXzeHDBwxInzYzQXzD5dymGGC+rpkKWUl+n5gQpypsSSZXFdWS2PtIRKMbve/XkNu2s\n5FhO1Rch0qWem2Y6nKataW/iZWPAgKiJrXesRHnv/B6/9/2RF1ZTKT2x64ScnGM9w6Uly0TwEvxf\ndyMyvkDDWbGU9yvES0Zxm8JBghB3upC4PnrANCqWjZGBWWoYtDTCvk5PgQeA4wGWqWWRNwzDgLUz\nyI77GUEf1c9X2eIwhKAJIZPRogA6JRUuOCkGogimwixPRHGaUIS3k/4+j/Jj0MCgArkBUZQOKb6R\nQGTUliDVGKSahLhBVONvroQmNBzlvdZRCaYapGIWYi8ZYSSbSF2HNY4N+3YByireEJot51wQTWks\nYJrxugIftPBh2NCkBc91XT/LZRXoNB7McLZxRj+OiMG2bdi2kUvrubceiqJc18vaDZUGyOCV7Emg\nVGUqCARPpDChlVQ97zI3j5b8Ab4YfFoq94eBP7cnVyKiIjIHcPc/+6oG9yrQ1l5Yd7jDHe5wh++P\nvYK1PDvnf/3Fj/nmWyd87d7ySx7Vb8Af/W0ef+RLHNfnA5tYz+alb01YgjhDF9mFDrXI2K24nh1z\n3WTMJsyN4JkwbEg1ZStjXOaJm/kR6+Mjct8BEZHSCDTPSuBrVsLdZtiBF6XC3dGwN7owPBnumRyE\nrTWHSiNqitGmpuBJNQ5vahi9mrZ0GNP8Id/pR65CIXZHOeIvlYw8HTPBrLoNwlF4i9FvAOGTfMMi\nHhUfiUpG3CEx0FYFIWvAzbEcmCLgRWVQB0wZ+4jREHDObKJzo/US4B7yF/UatcxkA5PDYp8Wq8Xp\nz4HNrJCmxhNeA0aANiuLpOx4QbE8d576jhdc89R3PJst0O1jdmJk5NAPCoQX6REv7D08FMMRc1jG\nE7r2nGu7xxDus2vKZHmtnzGBoWkwyzT5JeKDM6PWv2lfAmsJQMsUyz1p9mmaXjSDrXeIg4YjspVa\nOUVoNFYnyN88bJRpJEwZiwZeO4qF29d+V/4m37WPMG8Pc+zfeyyHKPsGu/s0vyqA7smeO32eM1D6\nj53nG749+wrgtN6yaVpexAWXqRg/PIknrLWjkX39Wonkx1Dq6h7F56hDJKK+t5930PZw0lYm1DND\nOGLTLg41T1pFy72DX9a9yYZzPPWc7q7pn/wicdgRdjtkcqKU9DfJ+UCm93YW5nIgHQCL3NDnBg+F\nUIrA3CsZp2yCzOIcpFhkqBq6l1o9V8c8IVjDfgpTDEQy91JmxRbomKp9aAgQrab07a9eFLRnOrQT\ndEorgUKiurAAh1aKEcatE2FCs1TN0gnoIRfVvVRYmTkJJWtLqGw+h5ZgDYqzbYsGlChptObG49jX\nGYZ8KMCsVDIrMUfm6SHXEWbXV4Qx7TVkXJzdyx+PV4xPS7D+N+Dl9sfz+tyPHLruHuN4p2Dd4Q53\nuMOnwc2zp4go7eqYX/zwip/8yumXPaTfAHf/F3+bx7/0ZY/vB8WLxx8yRC320tpjxV2cCcetBA+r\nNtLrlpwHkmYG61EbMN/R2hVGYHKYmhbvI02vzNI9jryYBaS+NILdR3dDjLQO4gmR0qS2XY8EN0Zv\nufEZ8w5SEmYpYpqZ7QNigat0SaoqhyCIxGIEAah1LENPNCPpVNzL6g52qd24Yl+34yLMxUkYoydG\nMvOaBgbOxHQISM92La6F/C9SMVpQiYWISENjtT8Rp6zZsfQt7omG8aCnlB/FCl2AyTP6Uv+lFTtm\nPmCqdDrRdAFFGZrSn0t8xxQjHT2bpi8dgmpq0zNu8BBpDNb9nJyFLbFQIBe0+LSR2R5GM9ExeotI\nuf85RH7t3infvnf7Odw0bUnflD3plP2ssy8mOrGLWv4WqTE0l4s5jtYqu2phsK/DiSua7j7CiuP2\nnKM4P9SDqTvbEDAiihClxVnSpJ4uFzodXXGFLsc6BEUt1tQ+IccFVq3t5KVQVElE2dJwQy+GKiQJ\nJV0SIQyJYB2tRZqcuQkzdn3PMp1wP5+xUyEAOxokJ1IWWjKR+FJKYUmxVZxM4myq4wgL1JzRRzzE\n/YygOEkCr49HBH85mbEei0IaRs23z1ngevtr7KYXxHFLJrJOMxb6Gus0IxMZPbI/2pz80hHLc8HL\npyDhIANNt6Bp2luyiRCq82TUDhFHpaYI58sDWevzrVK4Zksgc+4DkULyUoDUOVHKOTtTjrtvIBoR\nMTbpBVfpExLT4bwQIdym7e3vZCMtIhFDEQvovtDOXyIcZgRzopa6vSwdXntpqRjBS+2WSrFYH9MN\ng424cxhBm5Sb3TMQQYhMRJZ2TEc1wnEnOTQmhLZY/avDQn7I+mABvbsfuvPW3+evZkivFm17/64P\n1h3ucIc7fEpcP3vK4uSE955sGJLx+9784TblE5E/LCL/loj8+/vHlz2mHxTXv/B/lR1YL8G8154x\nj/WSSCSHTIOhoQSwz9MLtmaYQZuvCCQm7blOEznd1Hi3mCyU6p/AjNdo2gdACZ8eL1bczJzNAmY5\noQIxFyv2F7bkSo652R0TUy51NtFYWmDfIWltNzwanzB6LOGYtCBWU9gSjRgP87rUsNQ0xjHCc/mk\n1IjU5KQUqUFcoQKK8+yl4oZDyCzV2II5czsjeix1SmJEaTnxHcu0xCmmBEM/Z7JiZV0reg7HzBrY\nt+31ahH9co+tj8aPuNYF4oYGQWlJYUYOkW1seTZfMYaeNXOmcWJyreqW0BCY0eBSFAt/KQyT6oJ3\nGVY819JTqjSB1kMAa0BSJc1va60uu/3+tzP4jlsKsO9WBVkWpZeTF6VmMOF6ecZ3zu5x3XUQm6Jc\nUWq3gkRWcobQFQ3StdplG22GdSz1eH2KqAuRBZkZR9vaiJgS1M9yy8WwYnMIGR1CIcFTkLqWlX0T\nYaeh0xva+v7bAr6SPqljWRtWry54oKWhy5EXNFWrgCxCLtIQ5sY4PSfui9C4NXYYRXnKnGDOaY3e\nBx/xqlKZJ0y19Fur712ll0Ln4HWlVvWzQgFJ23INqfarkoYPtveYPB5UX8Xw2qNqT1bADhbqezUv\nS2koHCQc1qr4xKpZIigNOwzYxUIgcnWkrFTocNU7xkNz7tL1N5M0Yt/TRD5Ky3E4ruSxpAgOtqHR\n7jDHWys90VT8YL+fgGwTiBSF0g93vbSTABZjKiYmAurO1C/L90ZoiJKY+1GpYfRii5JTYvJcbfxv\nZ3hfqVV4f2m03AWlH0cgo24s2xl96A+3Run5ovBpCdZaRP6e/V9E5O8Ftr/N639o0bUXpPQCsy/M\nW+MOd7jDHX5kse+B9QsfXAHwjbMvbgfws0JE/nPgnwH+OOW/1D8GfOVLHdTngH79IbtYXPrUR4rx\nc0n7W/miWEm7MoYSxCfRWp9QdqTdYfIbtrHFJTDrAwtbY+K4wEqOQUqQ/mL6mN2uOtaJlJoqOpCW\nBROyD25Dw5QjYsIkGRw2AQZtX/K0cMxKUXwje3c/wd24ltobR26JizWAb8GcF8Mn3EjRD4YcWKeO\nWRzp1NlJRAxie1zPAuqZQAmEA0pCUev4tn9YCIOXFCFTAzXeyM8PtWdOIEskWeL6kGKnRF8T7Zrb\nIe4tFYwX7XFNzyvPbaxYlKdYgvesRRMykZpmFwjSoGmDeEmZEiLu+zHsNbTilOcEOvmY6C/Y5RVX\n+SETt+qFAGOsnZMcYpphPrH2NYMPlDq5osIEF444AwtcccM6XfHMn9GGstv/8dEJzxeRXd+xm/eM\nfWSuQg5a0/gczUpk5N5ly1tX8+KCOItIkOoUV65in+qWCSSJfJB/je/yHcyKGx/uuJa1qa412C+h\n6EhCig8/pUaprOeZCSq3pNuq3ftEx9d27/ITw+s0JET3dTh1PdXjihdHzNUU6fOteyHuDJXYZVfm\nY4NYaXR9e8+FUYRHNi+VSAKRBrwklFqTayodOLE8FJospLYE85vNh7/hMy04YntyT90g0Npryw9q\nrtd0wq1M7HxHVpgqES+UQojSgG/4zvEDhhAP9wIgponHu0/qfAiP8shWykZMwwQ07KKypSW5Vxv9\nOh7peCmBsd7dck8Cyra2DQAQ21uYgHtm0kg3jExW1MedRJI3pV9cNoLdki+LXSHFTVs2NkSKB0dN\nvQx1goz2t2wUbAg5R5rRaXa3nygJy6LkefmOSsdv/6bvfxX4tATrTwD/nYj8HyLyl4A/D/xrr25Y\nrw63vbDurNrvcIc73OH7YU+wvvXkhkUw/qc/+1/ysz/7s1/2sH4r/IPu/s8Dz939T1Oa2//Ylzym\nHxi/dtIxETARBoVJMw0TxkQXFqXOyAIhFXvwm75lanscJ+g+RcsRVaSZ8Ua75sxqnVS12x6k7FaP\nvsV94tpv+ECfQw3SS6inh7qRLIr7hErH5GVXfSSCVstkGgi36lSMPX3ckVs91IwkLQqGldw1Uk6Y\nGy4Rc619jQJ4YB60jFUUFWUnLSahal1SUvNywGsAZggX4+3OtQNb3VRXP2fpAyLCVbo6BNOCMNTN\nV6uBpTJhpIMlvIQ5+P/P3pvH+pZl912ftfY+5/ymO72pXk1dZbfbQ1u223E7GDmWnMRGJBAcJZgQ\nEDZ/JEDiAI6CojiMAqRYwQkRgwTEYCwIghATsEOUKERgJwI6HtO2u91dXfOrN935/qYz7L0Xf+z9\n+91b7e6uZ3dXVVvcJV29++79DftMv7vW+U6OiCeyg9klktRJzZrzvM2Qh+JSYxnhxW/NJ7bWDmm9\nXa+ZFAc+K3S+lM0rUP6vO1/H/Wc+yHx6NY9OGVwNdgBcOtgNlmitpmnucDwaA8I4KhUTahpa61j4\nDkSxkkl24oEq63zMwTquSSLEYcUkBnZtjKVAkIqXbj3FQ7I2MzSOSKLXWBA4Y5Wyq2VleRx6vISh\nr0ipJwqkom/aDDouKdPQsOh6or39Bs6d9Yw7iz1cGdtXsaajYel6DMeKCZLGDHUNGKf7T5GkJmc0\nFROFlI+eT8oo5H0bNKGW8Ja3eacXtPeYVpxpi3cTXFJ6EnOpMYzOsvdkozO8uC3SmuPNLlHDY73H\nPDzE6wSlJtpVfJTt9XN5nmVXvr4as67rbM8vWVrkgD36cgZnCu9mBsqU25zL1YVzllW23r9avu/R\nxeH2ut0Jt2kVjl3DnDFJZrzsT+i943ILLr/JV35CU4Vjc/MhP3ITBLD5Sdqc12bsMGZl67f9Pg9g\nidZXeWCTfBUk52mnN+ncCDWhrStWlceF5eWGFEpnHr/TdjszGp8/U+rBsszLKuo4cF5Xl7xjEgOe\nYb3ivaonDRr+WeBrgT8G/CvA15nZz7+bC/tS1j84W/CLFyvM7EoW1jVN8Lqu67qu651qcZoHrFcP\nl3zL7IJhGHjxxRff72V9vtowK1Yi8gyZsv/lZXn4m6h61TH2uwxS0Wdvc9a2ppG+tHjG4IyVDBzd\nHDMeDcxSX9AHKewqAVM0Rpx91p9+gwgMDirL1tNnnBNCy/HOAVLE+1Eqgrh8JztBJ44uKb1kz7KV\nJIJGKOjZkoqLrRWbIj6hWlrlBBHH4B3BaiwlztpTln1k1c25aM+3REEFbvUBR1Way02Qb7lzbcqQ\nPANCcpumL9O7pkW1kcThzRUjC8Acop7zcP62XVGt5+V1C8qAEC3QpxaRXZzeZKrPElM2jRDdwVxF\n70Kmu200bC4jdlcb6xQ7YpXvpFuMVwh8V451c05QsvmIOdrUMJ/cRlPF+WyHm7XhC+qFQOeFoA7B\nWKaeZTUmIfjoWdQjzpsRkcQk5rykqEZXCeKgtkAIPVhiJUtEMqpRhUAse2BBJAlbE4L5ZEJQx/Gk\nLtupBEukNBDNyLTKTPvzaUVb19zXmxuchof1nMGURTVkgw+pGeod1tWIVmaca8WmNTVggs8GCWQk\nyonyWNdEIpU51nhaacCNmY9nDNWooFi2YR3SSET0chAOLjHyK2Jxg3TJ8FGwsOYz8XXW0pH8iFVd\n0cWLbAtv2VhP+oANPZO+KchKGXSuCLM6TrMz5OYc1WwdrlLoaQW1y8Bf3q/pxi7nt3fZ0Dr7Gfuv\nggAAIABJREFUtAaBGwxMJOve3BCuDAzQ2ECwAcxo05yo40LxzHTaS1vzhBSTllkS1JRARSc1CaVO\nylBlDZ9awl0JzNPkGDf7MHmB6eQmo3rMXl2hqjmCoZyIyYGIbMf8/TTixnBJJw9Ot+PZ8WyX1ioG\nqxjKDZbOlHkcM6SmXMOlbONgKJgNiAUohiXbdy8RE7Kl8hpVFxm0hIgXirFYIq3fO/bab8QQ/luB\nbwR+G/CHReT73p0lfenr+375VX7Pz3+aP/fKA+rmNgD9tZPgdV3XdV3XF6yha+mWS2YHN3nlaMnz\n6QF3ZzPuHnz5GV2U+hsisg/8R8AvAK8B/8P7uqIvQT1rtzCEYIpRkYUlQkyFcgf0KlvkpGuyvfdC\nsoECBspNGh3hk24HLF+Gj8HB5v7yIEIXT7NGiJqLveepZeMpllsdLdSd+9Ud7rs8v57rikEG5ha2\nKNJaFGdp6wLndZLvQBu4mK2Ud9oKUsU8ecayT4xrunBBSImVxmwtLpEgjkRT9CACaSjjjyApC9rV\nIuZnxQlOiSkxsUBF4mabh9PKBeZujW39zoRRKCiCKiTLqILK9vcAQ8guaka2po6M8/4QT2wmBF8R\nHWzaqsE00xi3zWpunKMWNMISqWRAXXrPCcHVRBUWcsZZ2MFLw32mfGZoaLsJmbqo1KRMDbTcUKrB\nmo6EElXQtBlK8u9f2ZkxH+3RaeKTd2/y4OCpEiZtRBvYi9nEYhNuPJBDZKMovWa9SxJhPp4Q0wJT\noUkVG/2RIjQIhHbrptgW58muunTke0zLpzcDIpscsrxvOl/0WESSdURzDGm6vQ48gSQ+D8t4Bu+3\nyIgJ1ClTy4ZyaINoNiPxNTtyUPaHYJJb/Vr77bAS1bFQn4ebxrOYVJzs1XhySLElIQShtYDEgcZg\n5CpAeLRzk9PJHifpmPvpPmKXuWm9OFbu0sXPylpdWbgPZeCuK1IJyQ7W0bocatwzQAzUwxIfIlZ0\nZKk4bI7FYXHJuHqGqmtR0YJsGZridh3ZyS+i6TKMOJK1aibC0hqSlKviyl0BwXg5fYA5M1QUtMHw\nhaos2wcbFIfE/BzD8MW9MAlcjCaEKv8++BZDWOURO59D5ZpNXKK+NYl6WLAalQBtBNwE02a7OsoV\nZOX6UfG4PlMWLx0vhZRCwa4/N8Xw3agnDRr+74AfAX4HedD6VuCj7+K6vqT1337DV/CP3dzlP33j\nMfdjnqivjS6u67qu67q+cG0s2icHNzk+PqEaFjz90z/D8Y/92Pu8ss9dZvYfmNmZmf0EWXv1tWb2\nW97kQqPlFkI2gbSGp8rIinckB9ovOd17jpO9Z6mo8AQgMWhucE4lGzyoOHyfqTuuNJqZdqS8Nd5D\nigokN/w13nLTkn+WXb1GbcdIDqhtF0VIhWYoKVOyYnNz87KIGac6y9oKvbHdpt1uwtPzW8RhRbm/\njxNPSkOmBpqwkAHEaOKCoA5SzuXK1LLIOk7o2eVxc5tPu1skhLk1dH6MoizWb1D3ylMYT9kOnSZe\nm73JyfRe3sJCm5qGjKrVbkSjNwmMisNbdjXTviWdHXPmB87qwNxf0tgEoxeP+TGLeEzrzkvTb/gw\nEC0ylSk+5qEKDEsDDld8EuFkOGFDHBwK0vKx4Wnu9a8B0KsySA6qPZUxSQNiVqzqQWWMdznYtq8c\nF+NRaZ4TGx3PuZtRF3OJmBSpHKYFKRKhZifrxQTS6THTduMAZ5kOWvA8ANKae7Mxk7iD6aX5Aykb\nKNjG0p7IkRxg5nApm45I9ywrElGuZo11IFbQBnhtT/m1A0cIYOYZUmBd6JZRPIkphCmmjiEV4xcR\nxIyojqNJRoq6NGI9nrAvuzySERu660pqeioE8GngQXjIWX9EsoDiCN4V9El45Oui4YGl1SyT48I8\nbThHxLOQntfrBUc3DjjjlBODECaIZmuSx27GI327sUIU5bw9JqWAs0iiylRT2wwn+Zpt6eksZs5m\nrDj2t4h2mXOmKC6uScBbOwUt2uiabIOm5s+NgTdZ2CugFYIysSn3/DGf8Y85tx3aNMFRUcmAK7EB\nicijZgrOqAk0KePJbdqhDUKf8oDpUsK0jFuSVydlfXmfK7o13TCiwbJqSJYK+rT5tIALzaTJsRhj\nAj512zgGto/LutIeY+YvB/ChUiQlJjGwMb9oNZ93F/GCZFC7J43//eLrSRGsjwLfbmZ/3Mz+1fL1\nr72bC/tS1rftz/iRr3keFfiJo/zRd52FdV3XdV3X9YVrfpwHrL7Z4RaZSvXsMHDj+77//VzW5y0R\n+biI/FkR+aCZdWZ2/s7P+vIvTbldUGAyGJjSulOmBOpmyXo0MLYEjGhlnFsVLzjLzUrnKwZtiDLO\njWNpWKToT7ZdmR9xMDkg+YFYtEFKtkuO5BDPJkUOp0+RlGK1cUlzqpuGi/EeDw/2SAKehDX7PLYp\nJ2mHSLnXLNlKuw+XoTQhWwiQLIebTiywSo7DUHHELOs9DPZkk64FOmQUKP9VFwIVvSltUu7pBbGg\nVIZQyxQQzF+UXm6Nhkv3PTVwQ4u0A4kZV2/jt0ROfdaDJASKKH9TG5P3VyY1j6ucUzTz2fFMk0DK\npK0BTzIPcVn2g5W1X3XKy+88WAIxOorbHjn3Z/NIZ4VGtt0CcJbNHzpf0TY1ZztTkMgggagFpSPh\ndQCB9sbdK+eBY940zEcjbrY7KEavFbDc7olNI5xUC0rpCOroRrmZXNcjKjdBN/vdEuJrxgizlNcm\nZkSUNZ5BBDWQwrCLfkSobnA02eHlaeCB3ecwnNKnjr7O9u/oBGHELncZFa9BZzvbnRCv2J0/ciHT\nZAWWMctEokyLEyPM0wWt1fQ20MYlYIzTjN1wh8pGrKg4dCPaepehmZE0h1MfDucMtkISPNRzXpsl\nolda8ZzKiHmYcOwO8vsU9OpuuOD5/jTL7sTRDQuO5q9y0T0mpA6veaRzZISrtZYTd85Rc0LUgbNJ\nTbTEWiaIKEEuB96IMBR7diR7bWZ0tCj9LHE2OWKlF5gbcSKnPHKP6HTFUnLMgceRxiMaHaOWCK7C\n49nzB4x9x1giJgaqRKtYhKfQasZB8rhkhHEOj74zwMF671LXKA7TfI57G2hZZxRJFBNjVQ3bc3qn\nbdktweWbcy4bxWzMUIzW1yV2QViJ4PwmjgB8DDgMZ8Yq1bRWTDPK64lGUrhqh//u1pMOWL8C3H03\nF/Ju152m4nfe2OUnHs+p65vXGqzruq7ruq53qMVpHrDmOuGr4iPGqxUv/qF/BjebvsMz37f6fWSn\n4L8qIj8rIv+GiLx3tlHvUu3agjHZPXAaId+VXjPyx9jOQ6ZuyaIaMTZHMpeHgBA3/Tp9U2NuRhJH\nc+XPfkJIBsFqnDmSCYczx8mdMd1EQIWKGrWKYJkORjTenD3Fqj5jWS8QlFWssoOeOe4/8xUc7U15\ncCPfKTYd41OiryseT3cyfcsSTjy9bVRWsGROyzqzHxHe9AfctzGVRI5EeehuodZxR+7T0AFGG1sQ\nwZHY0wPqgjq9RLGiF7/t1MbkXKyvrpZ81eg1zndvMoxvs+tWTN0CTQkPVEPHTmxwlsNOGzoCicw4\nEuq6IZbuMdQjfPJbBG9QRVLM9u1kdClY4vZ6gxQq064n+lTojTkwNsLWLmBPmkzTcw1J4N5oj40p\nxsNOOO0yGiJmvBbuA8q6muX1xUsNTKoaUIeSaN3GPj/x2o27NDpQF/refHbA6c4+h75DgKAOh8fE\nEXRjJ67bJjZTAosNuih1ORa99yx2Dlg3I9Dc3EdxeAWpMpFVCr1t5TbIWnlVgeTH9KPbJNfksUDg\n1dtPbZV2l0YrxkKzQqxJGbUZXaGVxXIsfjk+4i0WWT8FHLg9NvKlDaXwsdxDgAtqBGGe5kypmbSO\nFcIqKT1K8I6kBfkQoSPRxyVH3Sd4+eaEzlcoWQuoBtXqNsEpVlVEn1GzqtDvVDMlEoFkkWV/kl/W\ne0b7EyiDkZrRup5E4uTGDlGFw3BedEyKmbCxpViNRtvjc3YjI7nOIl6MQKQ2gwrcNOuVkhi9BBJK\nJzURIzmhkh2m1QEqFUmEcX3AQhP33bP0emtL/aQgeh9oXiSkuNmjAOz3Hu8MlYgYdHRoUroijw1E\n+pQpotEbL9++ycoHBr0ykNhQtjLhw8B0cV4otcLaN/SSadJvd5YXNBk7ccATSUmYM8ZIpNSTcARJ\nrOfvXdLwkw5Yt4BPiMjfFpGf3Hy9mwv7UtaP3jvkv7l3yO+8scP9bsD8LfpriuB1Xdd1XdcXrA1F\n8Cg1PKUL7hwesv893/M+r+rzl5m9bmZ/3sy+BfjnyLrhV9/nZX3R5WRjEZCRhkBgzRlYRSyIx0pB\nCn3orYNnaMk6h3zHXEk6YqlFAl+oX1p0CzuxYZTGXDhloTULvwelAaduiECbRll7JMJk6DidHnEy\nOUTCwGj6gLXC1pzALjX/m8avrX2mzbkswPfiuLQdAE3Z+nugwpLQVdnifECZmEN6eHbxkAN9yL4+\nJgIvP32b470Ze7pEiDipAWWlCSz7BXpqBGMkI7xBI0rtAn0thGaKSiLYBC2oXoVgYlRuf+tumIc1\ny/oW9dyfZTQsVTUVWQ+12dplQRKSeJJVGMasC9v9YSkyuIjD0+CppabMzBmVM2WiYxweROlF8cXq\n3RA0etbWFjqWB4TeN6wNUoxEUyqraasd9qLwdHzI2k3YTJpRBKeX/m/Zul5ppeQymbEalqWZdll3\nw9VOttjPWwtpRR3P2NEhDyD1GCsDghNFVVgxAlW6lHBSM7hAU0AEswTisdmzUOVhaVUX3U7yrHQH\nTIhxzshWWekljk6zfkjTDer4AnU5p408IHYb90YBVduaZHi7PCsRttSzviAsKkKFR0h8yi34Fb/O\n6bTI9vj19ZgzSyxDS0iLzV5EJVF8QHisO6yTh0LB3GAsMSlNDDwzLOmcZzGdZP2aq9CiVVTgbLrP\n44PxpZZtOinXq5SbEtlxsirv1zvPSgaSV4Kvt9t3bo9J609TpcjgHb2vMIVBJnn6Rwh4Wmf0jedw\nslss7fO+UWAoWW1GplVudh9cEgFFLA9pohiJZu+C6V575bxxJEmIbvSC+bzNesGMRuX3EabDlbNN\nhEm3hBRw7ZreOy4VZZRpf4TlqxbFGEfHOIVy3QpDqlGteTRb8tLemzTr927AelIy4r/3bi7i3a5/\n51NvklT5kObT95wDRtcUweu6ruu6ri9Yi5Nj6vGYt85WUDvuzHb4E//7q/yx71S+6fn993t5n7NE\n5AVyFtYfIvdhf/r9XdEXX7E3UqUgDYGKi9BjXvHdGJlmJGRVATHnDy1llMNM1QHKED0H9THPN5/g\nYXyapILgMqIgFIuCjJJcOEe0qrjt5VZHxRVHQsdxbGhkwgBobHEhoD7SuBWwt+l/Sp6REa0DGrzE\n4qtmYJHLcF8jWSLQsOKQXj2VbXQ9AlZyt4aaKJ4Kw9HzysE+fYI96bkpp6gmTtIBL+sOFotDXWkA\nc7+ZGEfPUnK4bIXHl444kTBR+n7OBLhgRSM72+EwBnKOExDJWigv2bQ6pACpAypMoLIah2ftfBHV\nOy5EOEzKVDI1LQNgwkTXeBxBLi3WKRlRO7pL4xZ8kgtMK6wEpA4on9RzHtRn6M19nu5qENlSqAaU\nKQ1RoJFEYwOTdMopd8GM4+olXrY1e/YVbO01LGCW95iasTEUEPNE4laLBVmvNUjFII4hrTHJK2vr\nKTjH6/aIDy1m3G0GHpGI4nLulzmcOpzfDN2ljbbEXI0a4SJ6RGMZccpYZMJLt54iHoz5jnux0BZz\nfhsGLrUEmWzzmxIQXVa3ZdMXx1BP0MGYmvKWz+jJhbsghBLKm3LjHqqaZTPFqkDvA8ey4pZzZdPL\ntZAiQ1qj7Sk6mm4DehUj+IZqGJhX063tgkvQS0NMa5SaW8OSaMZiMsJiR3Iex5C3xTwxLsDv8NKz\nE776Mx1BHK/qKXdTnbVYcgvvJ/ihhaJ5TCIsrNvesFBRqqHHnT4gViOCA0ciqgNRTmVCT8vjjQdp\n2YZ11eSx2ozZasEsCjbe3X4O9fiSfHWFIGtQuxWnjTLpwIiIA+824yIsa4+6jpjgH45WdNTcG/Vc\nOLfVztUa2b844sGwZHSQkzVSgTdjQa8o23oUV1fePiEyQi0Sibw2W/DJncfcPd0DlJTyjHzRrGho\ncH7Be1VPatP+02Q3pqp8/7Nkh6Yv+zIzmhxFzUtJ2Ve4n3avEazruq7ruq53qPnxIbMbtzj/zCcB\neM3f5m/96kPeOHnvskR+IyUiHwP+Ovlv2/ea2W83s7/wPi/ri66FdgTNqE9UJbiGYDUxGkPwiBjB\nOQaEZ+wUb5FWa4J5jqu7NNJT60OiCPvVCSGuGLgMZN2UGGi9pvVN1kiwyWby7FjD3D7AMu0wpclD\nS+yppITZipSmFrrYMq/PaX0LcUWdBk7JGqaNv1obFtlUIQ48WH+KB7bmpdjSWk0UI7AJUwUh6y6W\n8QSNEbXAp6oDLmLNGo8bToh2wUUS6NYgiYFUsobydvQqfOzpPaABU55e3WIUPSTPfDhDUiKmHkQI\nFqjV43QnB+YW44mqz41wtFAa+uzAp5qICiB4q0ovbtREHtsF89JqdSFxbv7KluXHpQJfOb1JTsFq\nOK9G0Ac6NdqtEQA4UZJVtAKP9mY83t0n0/GyvmozCtkW9YRFUxFI/MK4I1GxcDDYsNXwYDnBaOMO\nmQfuiKA06zWC5aE8FbTUNwT1vKSvb5FVp0bDiMpluqk3zaGy6Uq+UgkkrlOhbpamecA4SpGE4pOW\n/cp2LQ9nT9HjMDH2wpg6bFwI4YbUNECD58Ky0+MrdUdslODzEN95o2tqREYMmphIx7ruSSb0MeMM\nb8guvzrZY23QJsdZ1dH7gfORlmPUgrUkcTnvK0EddTuMJqAfNax2JmB1QVgur7C6/3kmWnR8msOY\nZ+2K2o0YufXmVGAdTjhMxzxzuiRWibWreU2f4k17QCVaXDQjsZkSRQsKKQUBzeeU8xOmQ1/QxKxL\nywHhyqPxCY/GD+nE0ZdRdoOeOjaujFIMKgylpikmImsC96cT1sUNUKzC3IiL/RkXM0Xqnp48rHoS\n42aNSr6ZIpJ1jh0RTHjkh4xElmvrU77lZH4fs4L3adFfISSpyhCYj3TCWIxOOW8u6NOShMNLQ+XH\nVLvn7N64YNAAZjR9jw8hh2YjqH6ZabBE5I8Cfw34L8uPngX+13d4zvMi8n+KyCdE5FdF5F8vP78h\nIn9HRF4q/76rfr82JH7Xq5/kmcUSzJiHyGv9Lv1wtLW7vK7ruq7ruq5fXxeHh+zdvoO/9xkkJf5q\n/ww/+F0f4vd90zPv99I+X32fmf02M/thM3vl/V7Ml6xUSSSGpuF41xGdMlhNLzkXyiA3oClxJ17w\nXDoiJYodtVFJwEicuBE77oIFAwHl3F+SWM7TJLOhEBoXS3MD0TlUPJV4xlYx0xtE5YoeI1cyoadh\nSI61KoPrSdpRWaZobYXm5Wmx6CxOBmGIgRNbMBS0Y1St+Krpq2UF1ZYuJwhVtyaKlDvawjxWWzzk\nEZDvyA+s9Iwkly3OG7tTVBz/cO8FPrn7kWwoINm9rbOBdX/OjCmV1YgIZvn9eq0ZCtVr1OYMHXM1\nGAxEWhM2SjLbhs1qQQYTc2sZSss6WIW2gfMtSynvjCCGujHIARHPYjTmQ0cd0h1gLj+m21i6W6RN\nDVnRY4g5Km1QsqlJwyaMNb96MpcDmDEGyU12sBGGMLrizrbNFRPBPDR3XyI0c/brMUkiywrm43HJ\nbgKPJ9iqDHQOR1UwyUzPG1LFw/jc1lAFBFOXKYqmIDk4Owos3HybWeURxqm9pKGVIeWoMh5VgXtp\nfjlEXhlgKgYmaU1AWVrKodpakX0nM/3OWeShDdtBdEO1AzBROt9yJB3z0JBM8JrymslDjVhgEEdf\n5Ty4JmpGgssZuEdPpxU3h7y/z6RmWTLhXBK8DGWjPFGMvqpY1TXj7ZgKq9sHvPaBF7foKWS6nktT\nduobnIxndOo4Vyu71op7omyFfJvzd6gbvJtQ+aZcL8aiPqXTlpPqlLNNRN120FaCZAMVQ8BVYBV3\nS35XL561dgzFRdOnRBhPmE8beukxgTWeoJkyOZIB9QNBa9SME0YE8VThFvJZgdJVt6S70pPb+A4n\n2nPeHeXzybvtMRuLZ/Adi2pOoJjnGJwODzgbfYam7nlt75KlFnVaEHijdm9/33eznlSD9QPAtwMX\nAGb2EnDnHZ4TgD9lZh8Gvg34ARH5MPBngL9rZh8C/m75/7tWURI/8KM/wl/44R/CpUhUxy+nFzCL\nDMPJu/nW13Vd13Vdv6Xr4ugxu7fv4PoLdi7mnN1+jj/6HV/5fi/r85aZfer9XsO7Ue02eDNhIpi5\nQugr9KWCVuwWIXk2bIiIRVQ66nHYqKOIeMQyAvUBfZ1TqXjAPiupsfKqk+klbSggpOmUHKZa9A9D\nR0i5SY0knht2MOCoucXgNAfaSqaczdwujTZUOiOJ31LZet8wn4759Asf5GI8Y6gaNCamyyUHo1NE\njREda00ISkXDRLNORzUPjgbcdq+yrHrG2rGSESCs/AmdXtCXAVMs8XgsG2nKtiYS6H3Wgww+O/3t\nssvU719uvylDuevdhcfE5WNiNUIEhuQ5TzXz2GTntzIEqS9ZQ0pp7S9r3AeqYXWp2ioB0CJKh/BG\n+Ho+xjeTSYwJLUPb9vGysRA3ouQwWVXhdZnnhtt5vAnmBLXEKjjWCUBxdjlQm4BowixmiuTmHQRk\n7xyalnp6hkNQSZhmvUyiwocxQ12zGbOX7jaPq32OfU/nHeu04jjsc5ZuFKrc5oUFc9lC/xd1RRJY\nNQNLPy85SDA4T8DRptF2cAVwGL+2V/Hxuztl/Xblt3n7PGEbqLstEUZ0tKnidJiyLAYLW3xPlMkQ\nQBvOxue0/hRvIBJwwNoP9MTttQGZJrpBjerU4Mu1oGIF98vfb45ZwtArOXV6a0k1jvSTPRLKkBrA\nWMqEfmdK8hO+8vzWdjitQuRFXiDs3KWvKo6bhpkuWfuaxWQHQ4ipzwYWo5w1d4RDZY+pzpB6d7sv\niq8gF9bQNkeUMbxorjziZlDor0s8JxIwzWHJV4fwzXU08tMSOBAxG7bHZD2qwPJn13hoaNIBRoNb\nfRjijV83YDXFUTRZ2KJorfUsQ2ZLnDej7Xta3XC2u0e3d0AQw8xIJJLZ9gIPbpMuByY1YOzMLmi6\nU96retIBqzPbJsMhWVlpX+DxmNkDM/uF8v0c+CQZ+foe4MfLw34c+P2/0UX/Rsp7T/cHP8LkVs93\nf+zv42LgVT4IXGdhXdd1Xdd1fb7q1yvaxZzZjVuEkWdoje/96PNMm/cuR+S9LBH5x0XkUyLyGRH5\ndTf+RORfFJFDEfml8vVH3qu1RZepQKsqDz6dQcIzsqGgTrnVVTJNbPPn2RFR7XH1FZOFrRGFKzlK\n2TYbcxjKyJ3jK2V95ybLva8i4Lh/6w7ddAY4JjqllkQobUAy5cxmiBOq6ZKmikR8QToSg9QISsBI\nzU5GdiLMZ7s8vHkbs8TpbJ/ONVTR8DHgLbEh0o0NBMVEcVpjAkvXlL6wY928wcPJgnXjWLkJyQac\nZYLT4MC3F/j2glaWoBFnSh0zjWmDhJw3Yx7u7RKcgnqabs1gkfN+Sb91qDNCnGOnrxNGe/TjG1ww\n3e57U4Pqkn4kmlv4aBFEWE7ukgqaYUNbWITFfroMF1Gyhi7ZxklOaWIquNDla2tKOEtMbc3GyORc\nBhSj27/F/YMmD452iYykkpu1qSN3n1Z6khqG0oY8nO+lOXer+9kRUSPr8eRt52LTj3DSIEHLOqH3\nniWBczdwOq457h9xr3+djb4OYH9xmzpV3Fg9S9ggkAaBHgPWXIY7r3VjxZ3P4bJHEWCQbB5Q2xXz\nja2dXE5xW3pPK8qinhAlhykbifvscZzmnHLMkTukL/cnlmG+Hbiyekvwlt/Pp0s64loCL3OOmOWb\nG8KWTPdMewtsMwgLQWuSCRMbqOhpZY/zZkxXOZY0+JHixeOlIuBJmmMGwPAR9myXezeEN29Nim26\nsqo9yJjb9Sk7umDps6nLcjzKx9qMg5C3JA8pDudGUGh7ZQzM+00cHsEsIHGVt1MSp6MZ4qao5S2L\nKpj1pNSz0Dwsbk8qYKSu0A/ZDtOb3DDSwFDNkH6V6cd4vHjWOnwW/n1ZaqHcPoCmWxILlXWwErAu\nQvDZsr/YqdC4Md5N2AQLbwbhK0o/VI1Rc87xnc98nnf+0teTDlg/LSJ/FhiLyHcD/zPwU0/6JiLy\nIvDNwMeAp8zsQfnVQ+CpJ17tb6LW8zndN3yc1R9f852nf4/oPOv1hICnvza6uK7ruq7r+px1cZg/\nH8M6MDQ1h8y+nKmBX1SJiAP+c+D3AB8G/nBhXHx2/U9m9pHy9aPv3QKzqN+b34q9IeuuILeeA467\n6YJNS5Eb2JKzJI51M+PR7EZBQOBcelZkpy+xTAWqUgNVR+8qkipJRpjB450d8j3/GpGaD9QHhOix\nJKT6BR7LjD6MGApC1Ejg/m7NUDeoGGO9KJqqy0aY8orjtiXFFa/erXnr6Q9sW6MGWAwrMgaV/QBj\nabKLh1659650ztFVFY+mVQ4vTRvswJAQkJjRNF+viBGaUJccody8986TNAvuvR8T/JRz7xBqfHEF\nFByj6m7GJ1SIMiKlvA+MCirDmmG7fVqGpNlyjU8dc+e5mNxEgH4TPCtWhmdBGRCMOjmq5LMuRoTz\nWovtdiZt1ii3T454Id3jA/KIuhh3L6Z32A/KXghYVTFKHrXxZaspnyWJEDiu7nE0uZ/zucoxmVpP\nkhxmLS6h4kkWL5tWye6MbsgDYAQkXbbLyTIqFkK20JfUoRYR8zx/8RyYI6VIHdeYKJoinVVgOb8J\ngJio0kBnlz6TNZ5n5RBHpke9rXktg+S59rzkB87qmjdvPM95kwfgUJR/lSWaccubOw+OO0ehAAAg\nAElEQVQZnBK8p/cVi+k5i2pzxhjmmu1Lu1Q0SgqHLDjZWPe5y+HVLDGpztnzjxhoiOJ4TQLzQtf0\nusPCPc2iOmDejIhO6Uqm1vGtZzl86i7iI2gqZ7XHGDjeG3G+U+FIeJ5GN1EJSct5W70NTdKr11i5\neyDScCflz/KdUG1vtKzRPODHNUmEQ7dGpeL+JOvHaqnYY0T0CRnmnI4GjnYmW4TQENqUSaGGoqm5\n8qaGEpn2jmoQ6t4jZjTJ4RRGMnDb5bBvHyv8IKQUuBwLs7ZtupyDJo7iKQFhUeWoiTZREEVB3BgR\nx7Gf8XF/g9CfFh1XzsUSYDwI2yDryRnvVT3pgPVngEPgl4F/GfibwL/1JE8UkRnwE8APmtnF1d9Z\njlH/nEiYiPxLIvJzIvJzh4e/eaRpsrvL5OSH0DDl+e98mRurEz7w6CGf4muvEazruq7ruq7PUxdH\n+fPx9N5DAB7NnuEbn9t7P5f0jiUiExH5t0XkL5f/f0hE/skneOpvBz5jZq8Utsb/SGZbvK/1l/7U\nD/Jf/MkfwoUKMYcX5UxWvNG8gciGqiPbu8oqgOY7/9lzQehTS69jHh/sc2/vNp2NQDIKmci5NFYy\nsCZphNYdj0Y3GWvDnubhwsRRtytE8h1kEeGGNNQoiGcUGvphn2jZin3ReVpfESrPM+4eu3rMgZvz\nZrUxR0lZdi+Ojzzq+Jr1MyxGkSTZTCGJo60nWOVK5o0QVFnpFNQIaUyyEZ0mkgmt9wxFHHT1rjVA\nJCMQHRXtaCBIpCUy1EJbK73m5k9jxbiDpt6liRV9VaHiGekMBLzfwekUdWN8XDNk+X3WHdmly+Dm\nvV2MBF9Ttx0We3pVknqcJaLLVtrqAlIPeUgRyTlb4tnvCmpmhhNHENkK2PYiJN/jNTePSACDONql\nnjyHE8MhXEjPnfQAE9lmVW0QrGw8UIahgmp4qfAJNAqqI4Qa9T1DWjKkJSkFVqMRFT1eIrurFbN+\nzGxQxqHZpmWV/jovt6Aqk35F3RtdyvlkS1+TbEPTyxq2zobNzEITFmww2TebNeuw5FZ6yIEd8iE+\nAQixnPMfCi/zVeFlAFb0HLniMFhQsiDQMs6DE7A/XhIaLZocyeebwmJ8SVkbqoY65eFuFGsW1tHp\niLmrUOc4m93meHyA1AOTsJNRoGBb+3xDOa8qltZlWqDkga2zrgQfb7BTOJ+M+fgHnsMQ/OZ4SIWT\nmjn7zNJNAs/Q0zDqI4rgAyANpwe3We/dwRf7dEvweFKz9jmPaxQThuIk8E2LA75yvXF+zfsiSMIk\nMbgxn24ueG2czxOrp5Srm1HxJrwYD9w6fZ2bpzm7vcdzogNnkwmGo0ojEmNWoxGtG/NoegcTYRSz\nM+EwTKEdaLqOsfU4GchG7UIbe4hQh5it9DefaxoRTYgYQ8khO5wcc1QfQ7kpkCwPeStx3PVnNKOa\nYMKznLHrThFgEgo6KjBU7x0D40ldBJOZ/WUz+14z+6fL91+QIgggIhV5uPorZva/lB8/EpGny++f\nBj4njGRm/5WZfdTMPnr79u0n25rPUYsu8DfWf49fWj+HTeb8/ou/zuHBDX6l/Qa6awTruq7ruq7r\nc9b5YbYMOD/KIuPdr/+mrRXyl3H9GNAB/2j5/1vAf/gEz3sWePPK/++Vn312/UER+biI/DURef5z\nvdCX6uYgkClEVOXO/ohTXfJY53RVYBJH1CF7zrkUGRd8R/ZngLCejgE4tbfo3JLXRkdElEZbUtXg\nYmnKyiFtiGAOiXt0acxERzSS6U+dq5GYUAaGagwiNKpMpMEkFZpbbsQsE/pwkujqrHcI1nOsNYMk\nzuwcMeGYgJN9Dtwem7EkuFSGQs+n956hrUfb1xy8Y+kyOnDoG8bphOhWhUpoPOIWIFRJ0ViMP8KE\ntNGWxUREeaCfYhHPwBlv3Gk4n+wQLLJ3cc44FI0MQiNPsVxH6rOHPNvW7FXZtQ0RejvCpWGL+rzp\nOl6XRSGXKYdyzM7JMZO2Q8twlMQhVFsd24bgtm21LWNyWswf8iAyYjJkYwvFZytwgegus3wkeRoi\nlcAgjuQCljJtL1OtjDnT4nktyMYC3wTXZlpgQth1E7w2GEpDzczmmCkxbVC5SFJlLoEHsqD1FU4a\nfHiTpUR6zZTTyozBT7cYpYgxpEAQOHQVkmIZMDYamYKIFM0eCFEbxByKYxWEZIkzssHIVPN6jt0m\njDYH7hq5QR9Ju0VuDXjLLUkiqA6IDbSjKetqQpR9Rv0BsZge5L4+739Dqc2Q5KitZmEP+H9GK9Yu\nXw+HuzdJ4hj1Lbt9zUH4Wrp6xLrxzBn4hLas/YRBK6Ic0JVEJC81CKykZkm+PjX2uD5rvNo6u3cm\n0RJdMGJXxnyd3CQMLctmKENrzYW9lQ1AiqPfZrJ/OJvQjh1d4xm8o0plsBDHuW+Zck6leV/WqaEZ\nZlBosGJKMHjrxk1UPFj1NsqhHy5pqk6MWgf6qsJEqdOIdV0z35lyOL7JWme8dfAUu+sFqCOZojFb\nsvtxwE9X2Y1x64GSz/lxyAYlAMs0QZnls0TgkQ4ETZxV+XPlUdE8rsv6nnNniCZS6Gi6BbeqR2Bw\nPJsw+Oxm6fXLbMASkVdF5JXP/nqH5wjwXwOfNLO/eOVXPwl8f/n++4H/7Tez8CetWhI//uI/z5+7\n9W/yn6U/yUf3/wHHewfYr+xwtn70br71dV3XdV3Xb9m6OHyM8575akm17vjOb/jA+72kJ6kPmtmf\nBwYAM1vxdkDji6mfAl40s28E/g6XWuK31Zfq5iDAn/iRH+GP/Mf/PmfpnBPOOHELWrIpQ5V8QQkA\njK/pXwJy72rTmrg7AoE9O2Xu7pMkIhoz3SuucanoGMqdepXEniVG/R6kMYLyfOfYsUvPtY0GwlA2\nyU1aUC21iMS3WyCv/Ya0orzmK4yKgYGjcI7XgcYXe2oEwka34UmidD5bRsdyB92HhCUlWsXYLfhm\nPsnN6hNcpANACDh8QdxGfcuOndJwzCANQYSvfvUxawm4ULFzcZOlKMknhulT2PgmwW8oTnk9IsJQ\njxivIk0CEc9jJpzqiGfvPeTp44zs/pIueawDx64nFkFc5x2n/WPOzz/NLT8uBg7KtI9cKkSKSXoC\nJXLWX3AW4tsooB0VU6YkIoPl7V+r4cI4HwuUGS0v8CpqPW/tHGDtEg3D1n0vg0gesxqxPN8lIJRj\n9dTZMeKKQ5ttiJdCZT1v7Bzwi8/cyq+T8uPv65JWAuta+EizoOZZjlgUPRJMul3Qp7fngEiiS45z\nX2NuF1/WcGMYIwbR1+UIF5dEgZPoke2IpmhK2/PA5R22DfXdYHBfeb6kaQ4w6fGFNplt5wtlVoVf\nCAvcsIcmz6x7jipOQS5piIYWrVCk1oimCuhp9NLVUEgkGQhqtPWCi/oGeUwUKp8YJDI4yXlpyRjc\n5bHqUg8oP+/HhJCvsdtHD7hx7wELE4I6km6Om9KmhiPxoDCPLZryoCEISxyPTQmf9fEmwGsvvsjx\n3j6PpmOsauhVWNUzog7s6QU7aYQYjLs7jNoDxu2zVP1d+jji1VjRWkZbAdoKlqOGU9eTUBoZYRao\nXKYSjnnEiXuQESdxzIrRDqL0zlPFAGOhdw37ZLMJpwq+L8gqRFHaTUAyPo/LAugY0VHeJ2acxMCv\n+FPWzLdbPVhgLYq4Gt1YxEsgoiiJaMqv3j3g8c4432gZ1rxX9aQUwY8C31q+vgP4T4D//h2e8+3A\nvwD8riui4N8L/DDw3SLyEvBd5f/vWtV1ze998Arf+OZL/L/ud/AzzbfzAq8xPWn5tfn83Xzr67qu\n67qu37J1cXTI7o2brGpP6Izf/XVPkdaBs596meHR8v1e3uerXkQ2HQ0i8kEof8W/cL0FXEWknis/\n25aZHZvZ5rV+FPiWL365T1ZODRHFDE7Y5SPnG41INrXIbQWlyRWkdqhTlql5++uwwp08xnUBUkJM\nCGnC0m0ykYS4fCrrbCLUZtwNjgN2aPY+iBMh1DXnOOZpxaPZCbb/GBCqMHD38CEffv01hNw0Rdso\ngDygVFKDGMlK/o7POh4x5dteeoMPv/omD249j4gVatulKxlAV1e8fHN/u9anbUEhprFVZUlGhxrr\nMmeqNGtKNjDQ5IhDIg6B3aFhKqOsUZvsFXpifr/aEvVmiMSjVabHXljF3E04Dhc8lmGLNgH0Q4Xt\n3UQmz2PqiSRGlhGWu+cr9tsBE2WnmCpsnjeVBvAEyVvzIhnp6DUf8+hdeXw2bKjCKG+rKVNWxDSw\nP3+Ns/Ub6OKc4fwNNG5QrhzaKwZJIkES6+IaKCE37OYdum1ZLxU986q5RK0tcqxDHtSBabuDKays\nxqXA3fAKm0HCTLNuaYvUySVml+UwGBnVC+JoU08SY2DA6Di0KfNYENHakfQylUmA/TShCcrgWtbV\nkiiRw8mEIJ62WlITeJpTZmmFAVNpMWBez5h2d9iff0VB8gQnjl79FYt9UIvbM0qBQWqMnKPUVWcs\np2+wrnqOJ7sczfYx0ex+KIqJYy+OmLRrFn5C13isYNDRpmzNDwWSz8exigMnjKi7Dg15kN2ceUE8\nF80hFyNhcDmhrLUVS3bxdLzdLkKQlFiNpyynM5IKTjPFdj2qmCZhjDKiZiw1XeUzhdBqJvGA8yjl\nmFT0BdnTuODBwQ69y1iriDI4IViHAWtd0FYLIgkfA11x0NRYsfFRbItWtDLZIuYbsx2R7MoogE/V\n9twwQPwU8TcI1bScWgtiiniLzBtP65XWBqI4HtFzMcqoZhJhPLTUKRDEeCyHtNWSSMTSZ2kR38V6\nUorg8ZWvt8zsLwH/xDs85++bmZjZN14RBf/N8hq/28w+ZGbfZWbvqld66Hv+kZ/92/yBX/g/eOHk\nPj/JH+Dbh5/hlede4NVPvSPL8bqu67qu6/+XdXH4iFuuZrEz45QpO6vAw7/4cyz+7/t0L5+/38v7\nfPXvAn8LeF5E/go5CuRPP8Hzfhb4kIh8hYjUwD9LZltsa0NtL/VPkZ1x35PqdmA9yohV8DWjlJu9\nIApDDlc1wFRJanQuEkUYs8IQHk32ScBo3vHmzbuMTxbocMaQOgRHJBAlMbpQ6qVyW/Ld5FA6hEYi\n49EBlQreHG9IxYPacTLb5/HoFDXwIf89bYYBJasrDKP3LhsKmOHVsVRjsBHrcU0Yj9gEAU+1QRBO\n9vZAAlEcX59e5eNVbprauqJtHGexyZk/QBNrsvF0Hjz6rGYqcamwcZdbj2tUErX4bWBt7ypuhxlP\n2zRrRXRKFEeP0Y2mGJc0vIkuacqcbih9GPNGXPLAhTwI4Wms56vTfZxzjFUYKrgzP8tKETHqBDXC\nwk+YtIFQZY2Rljas9ndx4lBRJiKMtKH1FffHmzHvspHWpP8fe28ea9uWnfX9xphzrbXb095z29dV\nvap65XIZm3JPWbiLSUiwCUGQYCElFkJyQoQSgoT9D0GQEBK5/gCbIIhkE0AisnBEiDEuucCGyIWb\nuCrV2eV6Va9ed/vT7241c86RP+ba+9xXcfNMvffuczifdO6555y995lr7rmO5pjfN74PMc9AjYfV\nbe6lzzENJ1RhQcsRC2kR87zoY28xn58bNdFKpO0aUr+Rv+gZ22ZgA7bbEW3cznbvkui0N/aWglrz\nDCQUM2NZFLSTJappw3J6XTOB2ZBFigrf50iBbbLJkjvixfQqX/CHuYdM6j5ENzNUs5RZtdPJiCTC\nrXCP0Bdpk+goo5AkX0OSRIhCEEcXPAWRQgNDaxn6FiUxIGGuD73VtDYsZJBq2nIMj8guEUilA8ul\nfETzgYE45mXAENqi4HC6S9CL92VHrjGJT/C+hy/x3P1X6AJYatAYAUdKgkvKrPAsqwHBu2wAb8Z8\nOGL7zgO2X76DpMSv7d1DUJxFkrtLEnh+mOgssZJjZmQ3PpPe88KgS8JdmXISK3yMmw3+I6TopuCt\nq4JVoSwGA9rCU65XgnoQx6LK8tzOZTOORqGgJFgCD7Myr8oj9qgVOsnrzLnIiXstS9Q9WrtLri5f\nXW5tItI6LZgzXj/gImS6qnGuzQcqljs3d1cFw2ZFUxacDjznw5LOrcWwfWHqx5lRQ2hsRYwLrpUz\njJa5vvbQ6c3E65UIfuCRj68Tke8Hfld49bqiYNX9J8T6P+b3P/95ljKh8RWfevY5tn7xMgfrEpe4\nxCV+I5w/fIB2iqlyNr3J4Y99BhJc/TNfw+T3vT3dBM3sZ4D/CPjPgH8IfJ2Z/dzreF4A/kvgw+TC\n6cfN7DMi8pdF5Hv6h/1ZEfmMiHwC+LP973hrUEA98ECicn2m1brPpO3wsckW7UNPZ/lM+3x7SLuz\nQ3kw5OXJzf7k2Pjs3lMXG5E2MGJGKdnjr04dpx3kLVjo830MS4nTQcLUKMhsx2I4wLltFmlENh24\nMAkQDKeR+WDAwhcsCk/AgwkaK+qiQCiITkHzifJV2WY6zCxR4ZrN6zQCKp4wHJONGvpOHYN5saSU\ndlOkQD4hD73hRe09qSpIFTjfUbZAz+qFPiwYQH3FgW6xU4zpYoupw1JmaSXBkBNK+o2ey1uf1rV0\nEvucK+Er0ktctRNuxSF75nDe0bqSKUumLNlBGGrPcuS3D4dDzTENq0euIG8yd/0QZ9lRLWZh3KbQ\nkv46nUSa6ohoHVc45J32EiqGiiM4IYhn5hsqcb0pRJaErt//0DvZGcJKItOwjbNtgl2jKkqGkkh9\nfpP0JgGCI5hnKRXrninRL2UFDEzpCsdquoMgNIUDSwTniAKmHZGaB0XDg+kDurVphxkXqV85YBYS\nnWRpLFzs09eskwDRwKSkiR5wBMse7J7AgRzRxF0YCobgNeJcvk8UEM25ZriAiSFeMZW+r1BYai7Z\n10WkAee6RV14gl5IaL2VBHOvyeiS3gzFpxGNGYfdIXeHnvPR5JHZEuZusJH7PfPFzzDTlipV+JTt\nQzA4cgFftNwatNzTCV+Us3w9CZw4mpSzqpp+3Bj4tBGjvlZ+2vc5zYb5PtjTIa1BIQ4XKpoqS2/P\nti56JHNYQASBWV+Qhn5tCsKiHHBWvcC53kNSX7CqEoXMwAPaZ8qtos+FuhidL7ly2h/aWe7hLEKH\n9W6Ep2nJaZhzxBwwxi4HF5vBGTUdgWCKEQjiGLheCquORgWRwCo4xOLmcOatwOstkj70yP8D8CLw\nx9/w0bwJSNEoh2fUqytcv3eVm81dfqX8BgbVktG9lhAC3v+uqBUvcYlLXOItQWhblmenzCY5A+d9\ng1vEk5qD7/9q0u4AM3tbGV6IyAe+5FvrKJCnROSpdSbjbwUz+ymyQ+6j3/uLj/z/B4Ef/HLH+m+C\n0q14EK5y4oYkiZyOJkzmp6yKAYV2jM5OmVWCWblWxGHOsdjfRocXOT6GoLrxHETwTP2MZdrPG6S4\n4v7+mFG/JbNyibRlbqzfeRmpj5jMa4wGh4M0YN1Bs95kIVAMZrRizP0IcUsQw8mN3uJZeLhVkFO7\n8jlzLpyUQgtWReh7mbK0LvNbeZN/27c8uarZPq25e20bA0aSnQmTc1iCJJ5CHKEvHBBhaqdM4w6D\n6BitdjFp2O8c/XQhqrjoOCsihAZCgRdo3bqH50Ltsl1ucTueYJZoXUMAqtTbXxsgWea0665xJMd4\nEdQNKVFWZUETHNflVZQK4wo0HVVYUrRLxI8wM8Ztg+97g0Z0XE8v88n0LOrGCBEVKLpAWzpGzuFJ\nRMlGaV40SyxRGmxtX8AsKDF4KjdEbABpRdcXixPvuTr9VcYnTyFNSfItgzTnoqTJn8umpiNRpoS6\nxLwsN71Z63zfSETJsrT7u3skbSnpc7lido3LfE5GEmPuHPuWekknQOj7p/KaldRxKjvclId4hGpx\nJ6+4wVXWgrIBCaFgzRs4c4zYpqMhiKOKU/YimGTTnrKJVHXAjVMOXE4ps1hiBOcpmCEEous49DWB\nJe9KJ7xiq96XMpdOgYJo3cX91a9ZZO0yqZQhQFvjXKBxDUjufXN9oK8AQZRll9+7CCyrLfYomaoy\nSJF1VPiwrNFO+t8hfM4tcGG2Ftxhsg5Czi8cFCZuP8+XeCormMs6A65/kAknwTF1sB+VmBIFBQs/\nwIeOrnQsXKRNAbOCoJnhNoSz4pTDYsbVcCNL81LD+5sXmckTCHAc73K32GIYJsylY6orataGOEoS\nMFGePr4NlEjPMDmLmAjRlA7PWZghURhYTjUrU5bQ0hYcsWKqJ/jtUxY6JHZX8L3C20lADVIQitCQ\njc3fGryuysLMvv3NHsibBeeVn+8+wb5F3tF+M++59zI/9/Q38nX2C7x440k+/i9/lq//zu963MO8\nxCUucYm3Dc4Ps8PqKmbjgu+uh0x+301unzT87A9/gm/5Y+/muW+8/phH+Rp86Lf4mQHf8VYN5M3A\nMgzRBKaC+RF3dytudg1NIUxNuGnHBIxaBggwoMJpgXRLOlnHdqZ160df2GRc11e5l94BkDfpLp+Y\nlynSunXfjJAGFaAUTx5SpjGd6+hSQTK3YZDWZYgqiCmx9Gh1REzDzRl4UMU1Kygc5crTaSJZthY/\niGdQZgbCgBGea20/aDMW1rF7eEIRXS76yGYHaxXcdt0SUIZI3+MjtM5TqrETDilQVDrMYJttZlax\nTTaWSDgWpWPYxPwYTViqaXCPnPyvrzJL3fbODnHqWA5H4BQRRzMMjJr8+wdaMikqKCZo8xDfLHDi\nKFlQ0rFMJYN2Bji8Zf6iiB1Fu6TT/C7dSIeM0iFXmXLIO/B99lAVcoElmnuixHW0Ac7DHqlSTt05\no5T94TJTI5y3ipSOmnyN0+U5ISW2ioCkyNReZWFPQ/LU4iB5TFoKV9KpsJAFo5Q3rAOpOR2VTFez\nbFDST42JsJruMjnqGcjo0eIA2jNA6Poerg2X0jOSqxQoLOFE0dixXbfsrBpWk7WlOhTq8eK4ZZ/n\ni3YL62VpIWXhYjTBuQQB9ovElXbGqpzyObnBdM2K9azX/lFHbAJMgSLRyHmeo+0h22FdBuYCUDEm\ncsYNOyTqq1jaofPZTTOJ9n1lF2yaomAdEBDL/YdJlOOr11kejBACSpY0rjmxRVGx1391HJRjGfav\nJ0wpuSk7vErui1zpmKSKJDhxyp7kQnp9qGAI0Sl+lXKPnY1o4xJVIdoIR4Ejm3mYLICtfP8TOdTI\ndjI8npU4mtRu+hfvprtclXcSEUba0apjIDPmLq/dQMJCAkmsXMfQhJVP1C4xDokOI5hD1Na3dGar\neqdKE8OkJIt+DRXBS8JMUBPGTcu5CMvRaCN7fDh4wO5yj+3iGF+2SHcFhzDTK5CgrRxJcu/afBAJ\nbh0V8ebjdRVYIvLnfquff4lL4NsKTRf52Oj9nOoW71k+4AMvJngaxjbn48+9j9/7f/7kZYF1iUtc\n4hKP4PjObYoQWY4HUBvDgef86S3+2d/6FAdbJxwUDnj7FFi/mw8BXw9iuwMeVATpEs4CTeHpnNKl\nBgzGBGpZ21bnfq2xH9PIuncol1cpSc5V6suhe5zxPKdMbMyNwUV/QpFyxpMXJTnFqoLw9DOMm1sc\npCyZUomotZTtDiUTXFr1xVbiqz97JzfaDyswRa0PR20bbt19lTvXDpBpzdJHimhMbElFx9PxIctB\n8cjVK1uaOKPfEHJRyK036dEJ7+zusuQK51JAlTfZ564hqGwEZ9sPHnJvCkkUwQi43O9VOVKX+kLL\ncNaQ1OFTzT03ZpRmTPMk5pye/uxfU0uVHEuMESsMYT9mDqdBWHq45RpK88S0QGnBDfPINdG5V6m6\nI1TetblaH/L75dY8Y3RocvQ1BkWKbDUV8ppZyO+tD44zdimKSOtOIMFMlwzIvo8uFUSsd+WLpA1j\nJASWSDeCdgWa5YVRa5pywPF0QJUU0zmWIpoufCXVtXneFNQngtPshjfYQTDOQsk0xc17ELHNtYkJ\nHYqj2/TjRDEK9XzNnQfoqOALkynZvj4XGykZhURmgxVXJeU8rWTc62YYe7herjg24waR4/MJFR1J\n/Dqfd4MSgXSAyRzrpW9tVdBwTvOIGYIhDPryBWBAx3zUYdE/Sm727LHhkoINQVxmWciStWMtqQf3\nkWYCpkQSVbEkdrJ57Ui2M1lJ1bOPcL25Sm0528owUvIIDhcjFFCGxGo0paNlHVVsqjgzutRgBJZt\nw2m1YJAyg+MM1MLmqCXSUWvkoVfeKw416CyxjCUTDAcURE7SKXM/oXArtMv36SiNqGXJ2DzNaYNK\n4t6NM64vOk50RvQCTf7bND5b0A6VblQxWpwzCB11McpmLpojr2cY0GYGbsN1ClWIHCznvDo0hgkQ\n47Q6ZGt5BUuebjWEYb43O83mKHd3d9H+5lGB9Giv3ZuM34mL4H9OzgW5BXw/8AFy7T99c4b2xqD0\nCt9yi/jeKZ8bHdAulOvdXR5wnU+/+z1c/YVfxr7EWvYSl7jEJf5txsmdVxnWkdOdHQZpSv3+Pf75\nj/0Ku+4V/sj0z7PH5x/3EH9DiMhARP6ciPzvIvITIvJficjgcY/ry0XbjvNprybQBlO3cehzZgy6\nvFEyE/yGs8j9DgmYySiHBPeoR5HB1ikATfskzi2Ifr4xIJCgNDLELJ8iG2DJASWNKRUeMbhmxzzd\nGEVUqrjDKL0bZ4rFiCXHwSz3d0l9jUmdLZq3Tx4iQNXmUFM2WzxhLTTsLpyzGbcRiYnR4h6jrsGl\nlJm19Y7c8pbyheE9vir9El+1+lW2h4eYc8TRFMRnFz2BydmSW6/OenYib4RVIp3zGJFOYOInuf8G\nYSyn+F4nlxCIuSfrohMnj8FEiH2PyY6dQwyZ/3AtO8UJE1kRUVodoqnL1yyC+YBZpGDRe0FCGVpG\nLNi2c8SEm6++wrVXbm9YQh9BtOGpVQHiNg4G69G03lPLKS5GNAYSic8X93neG5UoRs5SEgwXu4v6\nQHrBmFlvTwJaLDCBh5MhjQSO171xvR5y7YQoZCaxLVc5WBcwHeGi6/kU0L7EEv7kRGMAACAASURB\nVJFc/ptt5vWZ7jYqFwWNtwub9mgFTmRjMpJE8OWIod9+TW1DzE6JhAlJYGoNVZtyAdf33c0G91hp\ng5gwTiUjt023M6ZVR5OmWS2ngmnaOHA+6U6YygxIvDLM1vxFlxmXl6sTlBbVQPJrkV7m55oqMzHW\nUzUGzE15UGf2pR6UFOU5IobTjtB3ct0tTlmW9wCj8Y7745KgF8V0INKRmeq6t3NZJSG5Iks++/m/\nv/1OZPAU5rITp7i1jDGLbgdiuGQUPYO4vuWyk2O+ivPunGHIssm10UlnLTPXcqc84974FDQyTBX/\nYnDAvKm5Yk9CgqCRZTrr+/YUHxJbsuLmySHvefUOCjx59yWuPrxPERo6IpZWdGLcVlj0vVpKorQR\nPkUka4BBhEaMYcrjin7FSgYUw0V/IxhJYra+GeyQ9IKxD+eP/HF5k/F6m4+eAD5gZjMAEflLwD81\nsz/5Zg3sjYKI8H3/8K8RkvJw+yr/auv3846z2/zS/lfzDvcCD7d2WfzKx5h8w9c/7qFe4hKXuMTb\nAsd3bqPlDqEouBGuc3z3F1jWT/MHv+l5ij/xSajetudqfw+YAT/cf/29wN8H/thjG9EbgGDaS32U\nwpdI17EYThjEFQsZUqhw3udirfeuSuLBaMrdqWcmnnG/BR+GBgEaL7huSRoknrOHvChPEJwDIu68\nZFrqRk5leCKOmCoKgYoR31I/xenwARL30V5+mFT7JnVyQWdQSaAGvMBwdgKmuLJmO54wuPcyaLmR\nAKp4CgYc+D1e7LfPW+mUER/nwel99l2BEyWQnddu3JtTF3C8P8AlJZ6/xDOLIz55MGC2NcA1BV1b\nADnQNIudIhqzgXThO5KtreBzqbCsSkhQdpHhcsWZZMYpn61fmARkriJ/njtotGAaoJwnBotDEnkv\neB4jMprDYoS5ki7l8iUC86rEFZ5dU1qLG8bCkZOSFcl26ylAbPMuWEoalxgFRykDroZI21ttILmf\nLRHYPl8wSEYzhXN/ziIazjeYrA1EEoO2JugEQ3ChpdXImU0oBSARNCGu5deH8MvDwCwNkHZFU/RO\ngQaLUOV+POgNSCzL/GybQV0RRzVDa7guR4h5ggqDsmNGSYgFMGXlRv2Y8vuw3UwRzhCJ2TBCBLWL\nAkzMcH2vjiAspaYwz9zdY2kFH+R5nk2nbDPkBX2uv3dCnyGVn+9NQR2xKFnaCCsLtCoY2oq9dsVy\nmCV6XdFxI9zjflGyJS7byRclsTOcD1znZcayluIJLuSKXAyiKqlYW8CvC+E8R4c7+xycnUPoUOfZ\nOzpFepYukihEOR+UDMhGLxM8SOK0S2jtCJII5DI3SA5CSJaLp2ipZ+yUKnY526ttcg7eeg4xBtIx\n1Y454NazH1sgvx+78xOus+RwcoubxTYDKszyvfP8ruddxxGHsR+u8GsDz1g8XgowuNU9JEXhwc4B\na8734PSEw3w34i1QFWeEZod33HmRlDI7ndKSsoU4zAVWVMcgVCgzUn/IJCI0zlN1lmWiatQucevh\nOXFQsZiUJFEWVWY+6TO9gvPcCTu/9R/bNxCvl8G6BrSPfN3233v7w4yvevoOX3vlVZ6Y3eYP3fkn\nXLm9IErBu5f3+JX3fiUv/eQ/fdyjvMQlLnGJtw1O7r66Mf/Z8ff59K9f4R1Xb3P9P/0rb+fiCuD9\nZvanzOxn+48/DXzl4x7Ul4vo8sYp0wq9o1nP3CRxaOzZgRA2LnsA9ydbm1IgQ6irAYKxsApMkGSU\nXQQSq3Eu0lo6kmQpTSKylJwZWXTbvBzGRE14EQayApcIGnoGJiFqDLq6Z0SMdnWLnOIUGC0ya1bu\nnrF9kEOITXKfy3r7OdJxf+qfr/eW3QdNlNJubLkrS1y7+4Ctecv2aaDoEs9+8Tj3AAFbZ0vEcm9M\nEX1feAD2iNOhxpzp5PLPvKRsQlD01tkp2w8EyRu/gBKB2LNsilD0zolVk9kBRalOaoZ1fm7nEic4\nQpiBdiRgqmMKcbRJ6UJJJwWKUrgsy8pzkmfjwI8QKzCB6WzBKiWmqw7UI6q8s73GMORg1kTCcxGs\nC0bOZzVMjORq6mqGFNnWv2wvMriNhMaWudxmEc8BI9IirKVVnqmMMMlBvNFN8Ob6EvNiva3NSLwW\nrHxFInJjNaXs5V6FC5yOBfU1WsDZ5IBhnDJu4V28RCEdECliCy4wN8224OYZhcXmdz13dEYURXsW\n81hPuZNajrPKjxt2iJdm0zvXubzmnSYKiWgCWZtoiDEdH3G8tUWcdlwLD5l0LWt2Mmi+J6IEonMg\nZMnhwKNqRJe5pzo6WgpGKV2Qq2LUZYEj0nglDptclFvfB6lKV25RkJjMliiJQM5iK2J+ZwzjdtFy\nOhyxvFpxcjRneXZEVKHoWZ7Q50whF/HM++fZzMOnXIY9qBa5d07vIiZZbGvCVucY9O+0kiiSsRYU\nW7GNJmMQjVEqIUVqq/nU6CG1awg9E3yBLA/GjBvdGfNqAAgDGYMrGDTt5tFP1A8RF8nnYZlt972T\npItKOzCiCPPpJtYQ8701R0jsr6YcNJ5xgFO9GIf2RjqFVJy7IrP+Ehm6MeIrluVbJ2h4vQXW3wN+\nSUT+Us9e/SK/SYr92w4ifPCbPsA3ftM1/ui7PsUWC6698hIDWzErK/7V134D3Uc+8paGj13iEpe4\nxNsVZsbxq6+gPm++nw6fpLUxX/O9/+5GfvI2xsdE5JvWX4jINwL/92MczxuCHVllqY0+0vsS8qZ+\nvalKBGLd8WB5qz/lNVwIOC0p+xPpdfiPJIM2UtQt24sc1ouB9kGdAD45hrGlrX8NPfwFGkKf2lQy\n9yegAfHZujxqomwjogmnniJ1ZJv3RyVoiaKe5xrRCa5UVDLz1cnaRDzLxjpL+JRwKI0opHVmU8Ik\n0kmDSXadcwSG7Zppy1vDjWu4GU+c3uSpxQiH4NMSiTmhq/FLutE5KgnXy6fEABcorn0C8bkZPkrE\nUuydM9NrtpP9hTGpc+6PxgQpIpodEn3ffjBbBAbNIoc7p4ZEJFgiRGNeDFERCuep1zJOyyXTNVqG\nPucDfaB+lStHSzSBE09TTmidoz6p2WqOmTZzChc2jobWX09Rh1zoasixtNqxdzqnWsw5tWyxv6zu\nkNQhg2yNvQ5PlvWcSBaNOdc79OEpTBEXNvNhAhRC7NmsxWAdVGt4cl5TtMQ6MUv6YjBpx25dYuaZ\nsqLWkoPziwidekM75bVbimQJYf/tIuRCeQ6skmPECjqIwYgxb6aTDvEodVX1r9W/c2aUml0ZcTGz\nrpJwFtEo67eCTjzmFzgFT2Igzeb3H/sFghJwzN2oF+D1ClbLn9ui5GQwwBlUXYFP0kvnIDbGMK2o\naIm9qyAi7CxW+b7WSFTj3thRFg21L/rbPssPoytJqhQJMJ/ZPRlR+3VGXF6j8xCIYtTktT7H6ICi\nF7J5XHZA7JnCzgzDUbFFVc8ghby+DSI1SQoE19c1ipEggabEVPYRjPPRFvR2GiL9eujfV2+RJxcP\nKOsjqi6wK+dU0mf6pf5eFqEZZEfK9ZWoGT46hjFR0qGpwBHotGPm2w3LXIQnMesdHCXhtMVQor7N\nerDM7L8Hvg846T++z8z+6ps5sDcSf/MTN/jRL7yPV/Sd/IHrzzM5e8izyy/ymfLdjHSKPzlm9fGP\nP+5hXuISl7jEY8fi9IT2aMFsZ4vtdsYXFt/G/lXHjecOHvfQXg++FvioiLwoIi8C/xr4ehH5lIh8\n8vEO7d8c26kAiXgCIoZI2JAyeT9heFq8g1m7S9FFIkJIjpoCsUTHRe9BoZGt2wvGxzWDNOfz15+h\nO3iCwsHUssRnt5sjsWC6WDKdr1hKg0lJ7Woqa5gyY6vdQ613Qus3wM52KUI+aTbbQiEHyJoQfR8B\n7ARNoQ8kzluitYwqOuXl7T285ZP3jg5WK0xgIudZ4qQgvqPo2Rll3cX1aPmTx9D5AXtNhVjiSryD\nT4nJquVk/JBEJGogdyrl1yqqE3zs8KNjoijeCRNTumFJ00/hIyRhvxE05paL2CL0hgkIRe8/MZzP\nmSz7oiFmid5oPmf3+JTos3Mej7w/lhpCXOLpwLL8y1WBippVuWQ5hOVwl+Ugm0s0WuBiyIG2IiAF\nIykAQ7tEkTxikb14m514zHS5QmJF129698Z3sd6MQfoIpZxvJNTS4GJL0Z3nxCMFkmHJOK3zg30v\nPHt29nHGdo7hcMim78r7E+a+xmmXWSVbN7IJPnq0m7AqOpzk6kdiXhNh2LKdcrBvwwCxyNNHp/gi\nF2w+dgxDw86iwUSZh4IPxs/0TGYipAGtGENLbOl8sz58MERyBtj26YJjv0Xq4wsCntU6uZc8D2Yl\n46LGabZBLy2SxNN5h7NEowlHselhNGDpi8xEipA0z5PGimk95GA57vuacuFg/SFJp7opjkdNu2Hs\nWsAVK3wv/TWMcrEClNZPcOMbDExBPGhJdB5UMck5agDP3FngJMtOa+jtYpTVsN6wXjvtkugUl2Ba\nzwm9ecnKtxDz4QgCC70DQF0NN8XqzN/brIOCAd0yEVKgM8/MBR7okpNwxNDVDGLTT68xXh2yOzvm\neDjlU8++m6OtET6lzC6LMYktVVScViTN91iVqs3z87/CYXXyyH2f79GueZLJouZhO8Q04cSYuLdZ\n0HCPEXBuZn8deFVE3vEmjekNRdcFPiWeL7DNT8m/w4en/z7v1Je5dXjIiezzgcUX+OzTz3L20z/9\nuId6iUtc4hKPHYcvfRFtS052d7kupxyGd/L+73x2k3tlZpjZb/Mqjw3/HvAO4Fv7j3f03/tDwHc/\nxnF9WfAmfCUF7w8zBpNTCjfGVJEUSS7LmfbbM0YqPByPGTYtnfN0q0C5aljaiHbh8ynz+mgdATxK\nw4gKIXI9ldxMI66EijM/gS5uNjK3/RknGJ12TDlnlJbsrLKTZGk1TeFzW44YVZu5rontQoK4fABh\njohjwIBiIKjROwuuiyM2aywHpnoE5YYcYmZ07tH+J0DlYgPzmuUom89ia+dBQ0kMw4pkwvaszYxN\n2xAH88zOhGwfvn5NNSWVO1RkFkSLMcejMahs5GVihqTIqmwJQRAxmrbhSr1ELDBMDanLDEglKwbN\nLAe1GhR1gy8TxVDZ8iXa1ZsriBu3RKOVxEAiyhmxgrPJMeazQchKjzDtctFN5hHENRSmePJ4nDl2\n7ByfWsy/wIG/R3Jwrruc+y2Cz+siud4R7trLhDIw0yGxZ9RcasjW2S130z7tI3Z8Mz/MuUsp0XUr\nrqRDlGzWsXZ9PNdcaS50Ts6HuoBawsUsdz2QYyZpjpCIoiR1VBaYuQnmwHU1VduiZjx37za7q9km\nKw0SRQhUsUJMs+jNsp3JQTxhyMX87j9sLn5/9wQJh+kCH0qW9S5NmmyWVN33m8W+3OksF8Lbac6D\n4TVMjGOf5Zb3h3uc7ky4v7/DF68/Rae5pAuPZK0aRt9Wxji07Kc5hWUGORXw0tWnSf0qTJKQmIv/\nggpvOTJAMNx8QV6GiW64s3l1p55p02Z2TLPMVlPuj5ucNCRJ6CP2CyYdToCU50STMJqf9D/LKoZh\n02yIv2SGGqy0JfWMZocjStuzckKZjLBK6HJOMEWccS4PEemoJiuK2NLRcjsMMRGStdy7cgURuL+/\nBZILXwP244xJcqRiB6cOlxJO0qb4NMnXtv6r1k8ckCMKtuUB0Vw+4HnEFOatwOsqsETkvwX+Ahch\niwXwD96sQb2RcE75+NM/yc888xP88t7/wwO5wv3993Lw8hEA870lP/vN/wEnP/3hS5ngJS5xiX/r\n8fDlF5kkoytL6vgU4oX39JlXn/3VT/O3/8YP8fwnfukxj/I3hpm9BJwD28D++sPMXup/9rsSg7iN\nmjJOyrP1i0Q8oazoUqTVltornZYEzfbjVhaZzQA0JIgFuqqQZGjqe6t6BmEtG1SMEQWoMYkFnx7v\ncywRsRHJAi01L5X3qXX5ms6bRiOdZmlS1EAsEjePs+edJ5JSZnM0ZRmdpGJTSA31oh+isLA5mc+b\nJWXGFJcElYiTtYF1olWlotyE8V6UVKBiDNsSZ4IzAYtEzU5v0uR+mfUWv6IjuS6bKOBIZtSD3Guk\nlm3xy+RYXL1GdJkLWIY6n/2HgEsBR+R97kUO7BRDqL1RDI65uVhQhcCDtmWuS5IuqbpTosQsjzMD\njTivNOJykbM2q7jQVTIuxhTFELTFyQpPR+kKqn67Evu+tLUkUC1QOKNCKaxlwoggLVfn8/w4Fayq\naRji1BHVZabFrO/9ysYBh8UOEiq8VBvKznUJb4GuL7DU4mbDGnsbfsW4mV7kSY4ZNLkPKkk+lCm3\nTyj9GWaWGUpLVN2KMnR4ywXFM3aXmxzRqieJ0JijiUtWAY4ocH2i8bheMk7NZhMrEhlSs1XPITpi\ns4/GiBKZW8OhneaerX5u5zLaLBpniZKG7fkNdpY30ViyZhTPBluQdvvLlM0hAuY4ayY0pgTNrNJo\n0bC3OGM1LHEITnOGl5CjuQEqCxtmD4Ep9SbC4HRnl7NySihzsTtdzNk7PSZhSEzcOMvzLaY0Bm1Z\n5OsXsrTQN+ymBaWxYcdSXFunCH7ZsBUV/KL/jqIW2bMjhMSoi4zanOemvZmFAK6pcaFnH/tcvdOi\n4cznvMSSLksA+7kdhCyv1V7aGwG3TMSiwmyXUVgBgpNIUsEkIBY3z5+YkdSRUFylbGMkXdGpwxOZ\nLs/69WM9n5f/HoXa57tb+nLLJdQvMElYMBblmIP67cdg/RHge4AFgJnd4W1uz76GYYTyOgi8vPVF\nPnrt/+Kwus6180N24glf2N3G+XcgDx+y+tjHHvdwL3GJS1ziseLe5z7Prs+eRqfLr+bq79mnKODD\nf/d/4n/78X9Ed/IK8vLPP+ZR/sYQkb8CfBL4G+Tw4Q8BP/RYB/UGwKJuiojsENgrwcQwSSxdwMzz\nKc3eUz6lfuNrJDGOi2OQiGmWbXUowbJE6mhrDwzGvj+1JgcOG8JLkwIjGxe0CVI3wFmBsy7L8sQw\nvySS8qbTCYkW9CK95szlpOAkiWAXPUwG2XGM3FA/sgYQJqczBr5lWC94+sXn2X/1LlVss7MeoOoo\n8b0Ir2fiLjw8QIXx6gbj+go+GiNbsM4YsrUTnQiekOfpkacWLofAiuVtG+YZ0jBmFwsRFzuGOujV\nbUYpgQ8Uz7PFAsUhpqgfXbymGtrlxv5OoKJhZJY3k4/QbomC6Gva3uXN+g1i7mgKlMU+LR1FCKg5\nTDyTTimSbSzTRSLOIuPYYHTgGrRnN8/9nGHM75mRY24BxEe8GmoORAlOwRcEEXQ5ouwShQTUdYgk\nqk4ppWNazzOjsbwJqz1Ss4RkOL2CWYFKYKA1hQbCuoATodKaodVcqxr24wnO+mJfNcvCtM374tBB\nzAVYBCbMM3OuSq15ZTlzjLkIjTW3ZHvykHExAzzS7aMYV9IJZxpoCWwxRy2RrOgDuGGSskRuUk8Q\n89kcBcPCpOc9HXtJ2EktRTRUjDtbz3F/972IOcCBJJIYk1XNsOd6coErjNt8sNCYo7DA0+kOV9LZ\nxsikkMyN+V4wC5k5MhFGbWaVGoFY32d48oCTgScKtGXJajDqQxmE5fYtZtsHIJm50qQQO9YdR9F8\nNsCJS0LMclVvkbH/Au/Sz/IkrzAJDe8+O2VLy77QN5xCdCvS8pyUchG8vcr33L3idDP/07C/+X+5\nLg+qfM967fBtpLAbiO2T5JmNwq/VBpeWeIt5dYpjWyJWJMQSYVCx2L6ZS6j+b9qgjVw5n7NzviJh\ndP19fRH4IGg+IeFzk1PQI8arAS56JjbkrcLrLbBas42uABEZv3lDemPh1PF37v9l/rvjD3EjPcXD\nwRGf2Ps4y+EuT89v81n3HO+dP+Du/i1O//E/ftzDvcQlLnGJx4rPfvoFFltbaDJ82OJrv/Uq//Jv\n/df86xeX/N6tB3zbd/8e9r7jTzzuYf5m+OPAs2b2bWb27f3HdzzuQX25KPQYNSVsPcwn+BIRpGeC\ncvkQwhCLWSpURWGQHEMNrMQo2xXoglOmfZiukET5Ik8QvSc1CepII112yesqAokQa8w3PFk9yc3T\ndxFJdNIimjeU0XKnTd5T5tP1JB1Bm02/y8qlfBrdZwHFouCC61hLfXJvi4mwlfLp+s7RMYM2S518\n020CVDc+09ZnUwnI5lUuCkuAaMKEeS5aEAaWDw68CY7QO4ytU39y5ld218s4G054ZryNlyGpDfjY\n0NIx0WzdfssdURcVjkSB53q8hdMRLoWeIZxk+RlgqaAQwRMvsqLIMsOGAUpgvJojKfZd/rkQ87T4\nGImyxKVcgDTOEVltbNHlEUZkHJZosdrIBgEmYUhu2suPz6xXnrWSgspGPRsAiHC3mubi3QRCpBjM\nKQbnSILS4Am5x4EdZmYrBaxZIu2cbpU32VF7K/1+bpd+ShiMYeB5j73U9xxJz0cac/nSxCBjKnkd\nJNZ9RxfGB2Yw8sJ2M+/Xv/UsCKiG3lxCGMaWqSzxNATpNouncIlC8/wMJZs+XFlMwGDSe0WGnHDF\n1XDENT2kspCZHbIr4Vhreq8KitSyjjRAOi5WILiUNhvtaxwzDUuGqcuZTiFQkJ0Fh7ZuS8vjOhsN\n+mBhaM0yi+wcVw++SHnrITq8yshtZXYQIw5HbGlmvlx/mBJIEPI1xzREmsBO6xhXh5Qk3ikvMZJz\njoanVDQ8ywtM9TaL2HAeViSBRcrSykF9zllwmWV1QySVpEey9b7irNoUOJuVVyVwiW13svmukDv0\npmlOEmF7MudEtjYTNkTw6W7u28RouikoJFkb8PR9elxIhAOpPxTJ60gBNPZrOlHKYmPRM4lvPxfB\nHxeRvw3siMifBj4C/C9v3rDeOKSYuHrviK+9P+Tv/Pp/wx89/k5eHL1KrO5x9cEpc9li8dRn+cgH\n/yRnP/XPSMvlb/+il7jEJS7x/0MsT0955+wFDg+u4MMWD0uoPvqDfPh4yMtPfpK/eu1X+N5P/3V+\n7s5HH/dQfzN8Gnjrgk7eItTDU4bTe8SRQHKIA0/J+d41suNDwlzkmfEh+z4w8CvO3RgtAg8r2J0t\nwAIrFptA05Ih17efpWJCNiqIjOIpuUxRvHU4jGjZ+7po8gnzOjR1zamt5U0A0hsFIIlRWluirwV/\nQjMYbayzSws0qc2bsr6oirbavJYPgUlsCAzyBlRyk7ogOClJvV09Bi7mHhwxY1H0Z/ZmlCEQidTV\nMBc8tsO4L6ByMZYQk409upihXX5+a0LbbyKtZwvz3k5yP0s9zzYQmjf9ZX+dzhKOFk/DlIfUriUC\nKRnTOGTb6t5pb70NN5L2hYStd9n5wgyjkAACPnRUbc37X/oC43pGeT6nbFuG5zXaxc01R5TwiAHA\n+pVKC31RGik6QaUlmmTGYLXiyfnFew+GFHOE2DsnCqqJUrKzY4VRkYtfM8FECanGh5g33WasXGZI\nx1ritCSV5abzxwBJrpdr6mYqEg1ec/E37IOF15t1H5U6bfEFd4OHdhVPTdldOMJ12uY1kKwvacGp\n0XBO9EtSsWDQpw0pMFCj0MA+ZxRWIJJJMxFjUR2TELZkydPhNksTXAOI4mZFzpsSpSunQMSFmkRC\nzZiwwMVs0IIIR5PtzXobpCYzaEkJJGrniSKUFnFa4dZpAhjLstxcfyoNpeXq5A477pjh9ikPDq5C\nAlHXF5hGCrn08Ckbz3RENGVDkUTBQpSDo8jSeZ61z/fjUkLvmONIRGt52M4wjDvdIcdhRbDAKi7Z\nX63vHQfmHl1iVE2HWO/21/9NiF5ptwpCdGyvZrl4NJhQcyOswIwtqYlmG8P3DkjLNh8eJShS7Hvs\n+kOQlBCsD0i4OKgobJUlh33p7k1Ia2mrCTuDGVV07D4SaP1m4/W6CP4Q8I+AnwCeA/6imf3wb/Uc\nEflREXkgIp9+5Ht7IvIzIvJ8/3n3yxn864Gp8Jmv//O88MEfYHXt4/yph/8hf+bke/ipax/j5sNM\nb778RGAansCWK85/5mfe7CFd4hKXuMTbEi995Of5ivkdTnd28O0u79v9KP/z/ZaPPPEv+FjxAt/1\n9HfxoW/9EN/65Lc+7qH+ZvgfgI+LyIdF5J+sPx73oL5chK1tLJZUoeBctzMb4yZU8wZSosGj5ZJr\no9tsFTNub2/zyt4ukjLJENPFZqh1xooC53dYlUPKtMvY7aKMqFKLJiN1hgs5zHcc8lamU2VLz5hq\nlhf1fAdt3weVSxzFWSI6GNt689tzD2asqhGV5TwpE/Bdg5LwEvvNuvFg6wp19Gjq2NFzxu0RrolM\nWUCfzeXKd1JUV0lkB7794xUqgoiy0GwKMD6LbJ/UiCTEb6GWMHF4c9ws8uZ1KfXGZe01DoRmuPKE\nLTfNhi5m2WI972gBo0qwJzNGIjwnN9mS8pHrhZIGs5Tlk2S+zqUVA19mN77+sWZZFtghnHdzlLjZ\nmOU5vRBVvqd5BYBbyxOKk1OeeemIwWnLzvkFk5O35BfslU+5uHq/+zWiBCwpkgYkD8N6iwQkO2Oq\nR2jfF2PWITHhpKYqFpvX9ApqQlBjvs4Uc0rnPcErVZ0LmA7LodPQl2s5P2rhCzBhGmcMU0uREojH\ndzXarJj5mpXP+Vv9skM1oi4h7ZTR2Ta1CA99iciFAUsngTC4h/oTpDVUFCfZLGAhLad+RekzEzNe\nrvvGhEJiNptoxzhVEoJJonELtusl+1V+zumyoToDt4gsRs8yWPZ5bFXu57E+e0n7w4euLNc+C5gU\nJGAxGm/e1GkX8KHJWWsGUZTVYEAg9TlO9po1MOwe4P0hA3dO0a4IIriuYRROURc3y7YxjxlI9Khl\nQ4ppuSB25yDGiorkimxK0q+V0N+30o1p4xVMrDdZSRShwzCO4hmmXT+2PKisCM7F8XAR2T2fUyw7\nBnWzWbElLX7YcViP2AknTLtz9usTCgLueJeRrQhWUvaFUb6SFohIBBcTaCvAawAAIABJREFU0xgY\nxH5dpxwh8OLN91AXw9wDmBJlEIZphR8kIp7BMiDkPLJWfTa+lEgRV9xKd3mr8KW87P8Hkr0sP2Jm\n3w78TqqPvwv8CDlDa40fAP65mf01EfmB/uu/8Dt4zd8xnAinv/hfcFQKt3c+yvtunfAHb38Xq2LG\nHf8x9ttv5vnxu3m/fZpfe/aDpB/9MXb+8B9+M4d0iUtc4hJvSyw+9qOc29V8St9OeOHaR/g/Do65\nNbzFj/yBH+Fdu+963EP87fC/Av8j8CkuxGS/67G19xzpzmcoyjlJFEfKOVAAIfFQ9khFgcaSK+MX\nsKDsr24T0phdaYha5PxP8qQkhF/ff5KrM5ikfTQlBn6KsYAITTKkVaIK0ve8BKcbxkic4fwKH6bI\nptQCVWMcW1aimB0QvKNMxqqXDK5iIqWOVHq0TQyn5zTtPqPeba7QMZV0GILr3dJK69AI+HwwviiF\n6Cq2kN5IfGNzkMfAM3Qu4Zq+x0VbIplREFkx0AGn4xzO25KIprSjFcLwkVN5w1uHHHye9uUxs6rE\ndx1F25JYb5SzBKnQbQTYSQfo/IKByxZvj4rF6POuLmzIs0Qw96dFM7puhrkhouk17MDa+GLXLTlM\nU25w0vfh5Qf5GJBkTPYcch57BiXDYdkMoujXjEHYlF/K2jgkSmAYPK0K4eyMplhy5dYXGURYnT+d\nCwFTnAgLLVhH09IXF7m3S2gDdKaM1LEn55zFBZ3P1x4VKtexG07w2lGaQ9EsiURwSah9R6GOTgQh\ncTPdRWNCbZsWGG2kZvkaFqkCSTylD9CwxPVSMAEslaTkQVqWrkEY46LRiGfQy+86dTgSI1uyourN\nLIxps8LMYcE4LXOZuyinDCgRsiRwRNNLBCOm0EhE1VDvs5RRYWc458HuAarCNmsrcaNsAmoBCZFP\nPv0sFCNkeZTNWSRlVnVdpHUB5wLmV2hrHJd7jJfnlDdO8cNEandxIZEs9gcAgrO+6PMJ2r7kFkGS\nIpaQGChEMYGxDXDdALU5SW1Tqo8SzBEK3SFxBP8ve28eZFl21/l9fuecu7w118rau1u9qdWtQdIg\noUEIEQFCgkEgBnsg8HggiIlgIhzgGNsTDkdgj2fGEeOxGdsYIhgjDzBgNoMFCA8IyZoB7aANqZHU\nq3qtrqqsqlxevu0u55yf/zj3ZTXYkkZSq1sS+Y3I7syqyvfOu/fmy/O7v9/38zUBg2WwbJj0skTU\nFOgfeRqEti3ptQvCMI33mQB97xnqEhk05HXb3VxQitmSuUT2whZy3HMMCcTTVPQOlTaD4eAa9UHy\nlhpV2twRxXBjbcz42i7FsiVqYFBYugoN2+V1xRD+ws+RaGAU5jxf+rwdLFUNQBSRtS/kgVX1PcD+\nX/rjN3MzoPiXgO/9Qh7zi9XorrfTP/e71Ien+JPPXORg86P8rd3vph0bzu3v8SD3UZx/O/ff9z3Y\nhx7i6gf/5PlY1olOdKITfcWoffIp7px8gCfP3gXRMBh8iF87tc+d9g7e+n1v/WoorgAWqvrTqvpH\nqvru1ccLvagvVedvuYA/3ARRihBwCN62IGkKoyjnIIrHYoOBjiDoFXIMwzBFXSJzNVrwzPoZntrc\n4f7zp2lifvw8S7GcvnFAv6lYiscG322blZIa1yhlu8naaM5geIMgEev7yWtzPGZnMAJ1keN8IKtD\nR0A8YtLCvKPVZWoYbxxQ6E1kNgrbcTUaBYgn05t3xAWlsR6Vmogh15t37x05ufTSHXwE191tl1Yp\nmn0UT1BDq77zrUAjbTe8qB1yvU9T9I9Lc2uWNNt7nbcM2ixHYtrgD3Sxasxhu3XkXQcnjSMqxvSO\nO2M9Wlw9R0ye7q6joAkl723kyB9hww2i3MwIAu3GF1Oe0VjnnGGP/qzCEAlhBawQegfN8bHohRTw\nXMQWp8nrZKzHacbM51jTjf0B3ljEebABF5VWLbk2OA1EjdhVIeMz2mgpaIiksGGxLZfNKVYcN6mX\nqE9gimACOZ6z/go2eJa9K2zmNeOsIdOAi4nUmB4rHXCDMqib7lwnOl0i7nUwDGAp0JoaJDC1JUvX\n8sR4jyELhtREFKMBow6NJgXnrkJ5u6DoKJZGXIJrAGfjNUYypdAWVYPRFBeskiEx9SGm5QbX1s7R\n2Jzuj46vy8b1MBpZ2JbK1kSbvIlGIhll8uHpzWI7o2VZDjj0ReqaAdZHjvJRIm86x+7pW2mzFfFO\nj/8bxdCKZV7meGsZ1nvglrSaUPLjWLKulix4BlgWRYkvcgSDlUBwafROVXEx0A99LjbnWJcc24Rj\nzxsSCbIqEpR+qOmOCmevTSAqmSZf1K4bMbUJPIMopqt0JEZM6IZnNb1uVPEYFMOWzlC9B5tfQI0S\nJEfFAoFRs2SrnlAzZ78H016Z/JYKiFDnFjoiqsSWXlQill6bjlljWo6yhmiUtiNtLrXHwNx8v/ty\n69/XgzUD/lxEfl5Efnr18UU832lVXfXnrgKnv4jH+IJlbvsUp+55mIvf+juM3vDTvG12nTrf5z/Z\n/TucnjxJJT0mdxVs7F/jaHCKB//FvyCE8Pkf+EQnOtGJvkbUvu2/pdq1XD53nrwd84t3vJed5hT/\n6nt+nkH2VcM1eq+I/Pci8o0i8tdXHy/0or5UDcsMhhkxmlTMxIgjY7Tcpx1BllUUW46SBvVt8sFI\n2rhVRtga75Gv19gUDEXjCnLpAYa5rHh/wqBWYhSyZk4kJhoZERcazu0fsn7DkI2vAjCoN4HkxXIm\nYmzDWcZYMvrFjOHWdaIRhkcJUF3QsF4uGd02RQEnkV5s6cU5C5+CS41Cvmgo2kQltKo4kzZxK+/W\nqmPlJB53zk672zmd3ckp9yIMhoacLOoxHpyjtQS1kJSZtD5No1/LzjPWRkNoI40K3gLh5iaf/hKy\nJaxfpS/L5EcL6XGMpo278571dkoudbdWwHmy4jasSY/V4FBN41dlkeNs8mENFkLQAS0NrTOYqMdD\nglFuDvsJAtM5RZilDbKpCYua3DddJ6wr9qKn8DUH8QFEFY9P3TZgKxzAtOgQ3AakILgMXXW1NI1h\ngRLVpDEwXOo+thkiiUD5bDx+S8bc9BEihxoIBMR6lgJT6aFEGrfAZDMwEWMiSyOkwbQuEywIoqnr\n6FSIIrTqGGg6Tx5APcEoraYN9A3OsbQZE1dTS0BMwDURV7WYroFXS0Q0HcfGjQiiFFUH+Og6wMUk\neQ1VhT4eDUK/GR1fa6pKkEiwWfLohdhBSlY9NKhsn9Z0ZaLGBLboipO8rlhkOdrBKCSk7/K2l/xC\nHaRlqSMaSYS7eT9BRkJxARPKBPBAjpuitUvHIBhBrVIOrtB2I5kljmK1MgF1luWwRy5CsDlLe7PA\nUIFMHaAUatMkLDB0JYKlsHOGLMhD6IorGC12U/EYYSNO0veIYDCoScc0RuHWgy22r06ofJ8mZhDH\nqK4CnIVUhvVZo0oLFaEgpUZE4yhGT5JzhceKG3zyVOTq9iZgcLF7v1q9H8SII2LzFqsGIRWaRW/B\nkVS0RkA8I10QpSH6r7wC67eB/wZ4D/DRZ3180Xo2lfD/TyLyoyLyERH5yPXr17+Up+J/fs9/zH/6\ngR/jF554Of1sySte93/yW4N3kNfrfOcsHYJHR/dQbv0u7/ymv8f2Aw/w7re97Ut6zhOd6EQn+mpR\neOrTFDd+l/uPXk5bKNeLPcqY889f/T+wtZbIYHvLPX71gV/lx//tjzNtpi/wij+rXgH8DeCf8TWE\naS/bmi1mRJTGTllR775+9xK3Hj6OOZOjLuISu41S6+S5yAbUwXUDP4JBaNzKFQSI8vGzN4dTSlMS\nEbxNuVAutKBCTrPqpyAmmdnTGJawoSMQ2MiugCzJTUl/fReKFowStUoQjtByengJSJ6cLAMTU7dh\n1alxCLk6pKNOdB71VOytHA2SXr2IIYjBYMmkI4Ol1NEO1BAJBA6cp3I95nlJ+nVvkNhy9uqMGJRF\nzJnMxzQYvIe8VpbNatSvGzscXMH053gpWFEHM98e0wABelLhukJGNGHTUxxw+jetCWn8qp6iIgy7\nm7gjdxEfFa8zDoqMJi6IdlXAdGdKFeNhYQxBYbV9qsmJOFzn/TkGVGgkypJr8jhX7cOQz4lYTk0D\nO4cZZ5f3UCw3MSGd+zk5M82ImgKQ83mDESWPnq3lHDQwKvdwzqN2xWRcJS0pGIONEWIgaECrCepr\nQtd/2ome0qQJrqlULPBIjGws56xXC/ImUFTJdyNd57EXG4xGTMxRgamdspdPnnVEFUj+psyvExtD\nOfMYH9iSmlYNUSxWGiIGY9L17mIH7Zdpd21B0WyCGjITWWCZFpMUBnw8pifUtkfhfffz6Lrj/awC\nOJLGAhGcpA7QoiyPCzztOlXSpO+4a7HAEI6jAyIWE48QlCELrEZGYlAJPBtEYchQU/KsCUJElbKu\nuhw0w8VJTYFwOBqlGy1G6MuUmQxpEE43I05X4+S5w4INDOK0e3xJJbUIrvT0mSfIhEJjqzTgtxpd\n/AtnQyB0ps9a6MUSFNysQiZKFraxBPLVHB+WIDcDnVVch6pX9oc9JE5AJogqjUkwnNgVu+kbEjVS\nY6QkZbx5BtCNPWb9GUjL2lHNyLcJSU/Ntfb5S5j6nB4sEblFVZ9S1V/6XP/uC9CuiJxV1Ssicha4\n9tn+oaq+BXgLwCtf+crPWoj9+6h/4X7i0ZLDpzLe/thdvOY1j/L1X/8u3vexTV5z4w1szw74RP+v\n860v+UMmH7xCwDL5pV/m4Ze+lLvvvvtLeeoTnehEJ/rKlirxt/8h88OSR27/BtDIA9t/xqtn38er\nXvYqps2Uf/mJf8lvPPgbtLHlltEtXJ5d5sWbL36hV/7/UecV/pqTWIvxGSMWXDse2kk5NtYKC8no\ni+82a4rxCw7yW4masb6Ys2EOmNsRro1MtjIuV9sMSZupFb1bUD4xWuNFsyWPbw2Q6JlbUFFkPCEb\nLJCjTYwkD4QBimCJKNG2bLU5KhVBHE7TZrNQQ9DriFlL6+qCRysManIgdqhnJZSG3qzGyYDTYcyc\nKcoRf9FEIfTNGnViTmC7Im/lm4FNjNzoSO0J0RwEPnDXbZyRHveufYrsWkTGC+xwyWA2Zi9f52LV\n4iKEvCYbPUO9WG1mXeoCaIM3BusaPBkpmBiyKlA6jyvGiNYd5pzO1ZRkO96ZNTVqc5wzeIEQcyBl\nMK3IjCoR2kto7258ZjuqnDBYzlhfNhzaBii7I9Lh9olo6xgWc6p4igEtkxAZ6oKlgaC9VJAHQ21i\n2j77HqfikEaT/2neFlStgFsSWofGzgOmgYyGnfETHDUFPhrMcv24AL5sdyDKzRHLldfPNxBqKLZw\nJPy9Wk/R1jyqjp7dJdcSjcIpd43V8FyvDsS+61pxNzfueT2gNUtWWHc0jbEZsUgXFXClFJbacLFx\nIAUSI3OTcPkGODSxA56QigCjlPIMC3/rX/hZy+jG5AS24j4QEJMw4OIrhAyDwVtJPh+xnadpNaKp\nrFDriqFfG/YLwR+ftfTYpW3IwjJ5/TQtshVLEVL61rCuWfS34NCSacMsHzJGaF3KphMbKTRQWsPS\nN+T1gp6z5Jnw0qrhRhbpZUfs50uiBCpbUbHJ9aJgHFsIjkNzxE6zg5GGteDY615BAnREXPS0FOQ0\nNOKZ9feZNhnBDJN3zvcJ0SGA1wolo6Ymb9xx+2aoM7LYw2rEaKRXNLStJWBpZQfpYDhit5gMNlhb\nzkGPwEcWNOCXZKXi7CkeuyVyy8G17ufKE6UmGMNh3rKG0oiD/TX09BGDmcdozZndBlfMsWtwVJc8\nMDvNd3+O99rnUp+vg3UcDCUib30Onu/3gB/uPv9h4HlpE735vYEf/Ijl7zwifLvf4o8fOM0TlaX8\nut/i6uAJXn5jxiNyN3KL0Hfv5mff+A+449FH+c1//X/w4Yeefj6WeKITnehEL4jix36D7PD9/N6V\nb2dvO6fVyL0HL+WbXvvdfHrv03z///39/Mqnf4U33f4m3vbmt/H73/f7X5HF1Uoi8l0i8l+KyD9a\nfbzQa/pStX3nWQSh5wvGYUJZ15y98TSuNgSvNLZG64a2rrF+QZDIuckzmLzhTD0hb1vu232IxpXM\ns4LW9lAsglBWddpAkiJzHthZA1b+mFQAmWGHZx7OmJmKEgOmYW3Sot7imjlVPaBVdzw+lhHYcBWG\nCsn3GcqQnrg0+keL70YYccpkG6oNh+bgLUh0LOlygLqCJaPGaaSQIaKRloy1bcf2+SUjv8q2ybCS\nMN0rG5MiBGkoe3NaGdPbWGD7iYw3bpecOtjjzI2riJ8xLh+nX6+6V2lDHDE0HSFwePqxBDbQBtcE\nRkcVo4Ml/fkSV/nuzr5w1UWm3czeqfKedOwEMmp2zJPkeC6zw3VZp+nuc2fdc1zv97puy+rsCwHp\nPDEtKOQ+HHcRBMFHmzon7WlGvqCUChNDB3xIOUF1ns5hXyasTQ+5uIgUEilnntGiJvr0WqOxeGex\n/SUbV28wnBwcdynK0FDGFtTiJRCOcRJgY0BlSRxWyfMEmMxjpKEIFcYEnnLLVOzZOSoRJRDtsnsd\nYGLbofjh+CAoaEwQlFZuFikRYY/LPO2eBhRVmNqMBiVqCzFyzVV4PFkLrk3nwGqkbALnmgmOKSGS\ncN6awq43mHDkPBmRjXgdfI1pPZkP0O7Q1j0qscddRIsgUTnIl4B09EQ9znWb9fqp2+XS+d2Khzgz\noR+W9ELkse2LrO4PFKFhvp4z9ELfD1irK5xrmeaGw/lecu5pKsTL8RQris9yss5zV9ZLHEqmkYJ0\nHpYuMMlbgqQg4/JoQabaXRl05zX5LL0k6Ik3LdLhcCIpz2w12Kc253DUp5CICxlX3DWiQM+UNMYx\ny9apjEOCYyQlLzl4nB27l3LzJKR8vSIy7eWsxgVXB0BViQSO8kOoKpquu51JAu4cjUccjja6HDUI\n4oguY7y1iwLB9JnMdxgFWM7X6S989xOilHWkvxzis5tB4F9ufb4CS571+e1fyAOLyK8DHwReLCKX\nROTvAf8c+HYReQR4fff1l1Wqylr2MvLiDezZ1/D0pVO88Z1j8ret80wrfPDF/4o3Xks5Dh+PX8/5\nVz7DbfM/5V0veiX3fvJ+fvlXf51/8rY/Z9mceLJOdKITfY1pfgPe/l/xvng3k+3XEV1FsJ5P+nuZ\n9j/M3/2Dv0sbW375O3+Zf/pN/5Tb17+gXwPPu0TkfwN+APhx0u+vvw3c+jm/6atArsiobUvfRLJs\nyosmN7h14wl0O+0aTesJPo2TOb+gNS07coVX1x9ip78PGmmcwQZL1o4I2QaVUbzR483z8WaHdOcd\nTRseRehrnjbGRoi0zO0CNPmz8iZyYXaa9XaAqkcJOCyWhMAOGEp6DM0GGzayyR6FOBpJ+VvFmatg\nLGWo6LVz9pwhzOYEEUrfdF0C7bxPvkNhK94I5XCAFUPvWSN1RpS8QyaaCF4tGlM4LSK0RYsj4GJg\n0wgb84Q4R1vQmoNig4DgOriHSsHK9zWJB6yZJaflBkU2ZtQfcarYZDBrWCx6eJPRSkKzz4x2OTxp\njE5EyWWJjyl6OIphqkV31GvqWKFUGAJT8yircC7jaxYmS51Gowkc0gQIS0KXNzQJPZaNS9AMY8hp\nOCt3cD7cmzK4VHmqWGA8lAvP5vwyd1x9lH4xZ7DspYKoHdLGklVhaYzShA7h3Y0zDmKdBk0l4E3i\nBkYiaEBVGZ6/wnD8NFLuI26KIlwt5hgVbEyeudBtmjERCAQbOkBKKlg0CrlE8naCUcUGz9w1GCJN\nd6kGoJGAqmdulngJ9FkQRcm0YuvwgI3lElCK2AIGq0LWRvJQ4bxH85pS0+hruYzYCFtVgRNPNJGQ\nw2IhhFnKIesvBqlQijlLSWOxkcjAF8lRpOm4BRF8d5NhNuih1qAmhYIrMDeBovVksaWnDQf9Md3F\nixKpC0e0Q4ZtTJ4rs0Q05/DCbcTYefMwGJe8ZLceXuqgHoqJMIqBGKruukvPmVPw/lvvpNQzjOqG\ntb2KgYe+Xz11RARc09Dbn0GY0tcpcLPSj0RiViJZiQioyRGFiZ3iQnrtCyfUTmg0R7AMzBBtI+P2\nBkHTTRwcxKyk79YQDFGEg8Gog1gIwUQK5ixcYCA5h9mYvIsE6GvGtFesDhfRWIwJZHkXD+D7VJmh\n9SNQQ38xOr6WjcKWz3ly/dTnf8N9jvT5Ciz9LJ9/XqnqD6rqWVXNVPWCqv68qu6p6rep6l2q+npV\n/cuUwedcIgL3jOH0Fnb4cvLhd3Pjtv+cZf0jDH7pr0F5ncvD32FY17x/+XrWbp1zavMBnrzzDpYT\nuP3oMo985I/5zp96Nx96/Mu+3BOd6EQnen6kiv7Oj3MtVPzxwQ9ztD7FhMiubiB3v5+f/Og/41Vn\nXsVvffdv8fKdl7/Qq/331WtU9YeAA1X9J8A3Al/1c97GWh7rX2ESaioEd/4SBghrlokZkVdpxEiM\nQbrITgFaiUg0K2YdQXKczxCXEUQJWUCwBJFjY8Usa2iMRyL4aKljiZfUZRFg42CJqRomZooMltje\nEYNmSIZjEHsYPKX0EkreGEKxw5YMGG0+Qqk1RpW8o/cZU7LCD2Yh+VjoMpO0CGACViLDkDb23ho0\nGloLbdHiRcmNpdqssKNDbFTyEOiHtOk3nSdKSQb/R8YX0vFEsRro9VcEw66Eyjwqjso4Cq3puGgo\nbef2aVAkfS4FA3cOK/64iwNQi2ey4brXpcSV2zxGnDTIfM6sTuGv57NzJIRHQAmIr7nVP02mc5Dk\n8IkdjD5gyaTGdkS9wmekba8kvw4jIuCK1DHJtKBwqbCQKIzdnJApvXnKiQLoZxUZWTdC6IjRoisy\nY3CIKEVsjkcYQahNTaTFxECDw7UzGrVYd5DWFjyCRW1D1EDejYuC0kraKPcmnt40deFUK0Q7P5Qq\nWTwgi3OyeIgLgdFyFQ68yl1TvEBhlgQRbOx1kAVlxBFrcR+nBheUlkDmWwSwNrI5EwquMGifotqJ\n9H3qng3nnuHcUzSpqD6b7xP6Vzg1P+SwN6Z2JVU+SgW8bRO+fXXCFXamT7GxuHZMmDQrgEaIiE0F\ndtVh8oMKprMqjcy8e4gVGJ3uKjMEViOHgaONcwyKiil9cl+hgLeWlx08hKU9vvpedekqr/n0w+l4\nxZBGFxECkdYMsVpScwGadP2fn/XIfORjW7dxUOYU85rKL1jGCVU4QK1hOeiyvhIDJRVss4aerjGI\nBamXKKx8eW3TMM1uglk0j5yVCRc4SP7IUunJOt6W9MwYX1jcsMXFSBSDz5QxCeufhwyV7uYO0CMV\nrjZErHii8WydvoLrAqeTRxK8XeDGM4a37nbrSGHekO6jPF/6fAXWy0TkSESmwNd1nx+JyFREjp6P\nBT4Xeuo1T/KO1/wy73n9/8KnXv2/c+3cv6Pun2ex9WNkv/dfcFRc4SUHN3iovIvFco0Lr93ltumf\n8uvf+B1sX11yp7nBrc0T/MBbPshPvethYvyCas0TnehEJ/rK0wd+hiufeTf/6/InGMgZ2mLCImv5\n8Lm3czX8CT/28h/jZ1//s2yUX/Y8+OdSq/muhYicA1rg7Au4nudMV0aKFw8aCWLwmaMq+gTjiDZR\n7VKjR45hCwBRG1QV7wZY00EY3Dp0MGVxgYYcr4LRBcbXVLZJ3quNXZDAw+UdXC020+ZRIZt6+pMF\n2Xif0WiPYniZQmtU0obedh0kNRHEwuZ16KUtQ0SJzhCdYWxT9eFUaMV01GWlkozSeuK6AYndCFSk\nb2qyC09hXENv6yrRQp8dQi9Q9hvEKX6tT9pSLcjCNYymjCKMYeKGiYwXDKI5rTZgEqHNxYjYFkvA\nxpAyhYCoaWwqcc9SntYgO8WZ4QWMCpkoFkcw9XG47GeyXbwzHaIibbZLUtdPq5be3kEXH5WBRpQe\nqCKhIipEB7m0iERCXBCB2qaso64RR9ssKJY1NeeIxUtw9JDW8GQ5QzZaRr3AORuppUc+q7oOHgQn\nKBm58YhGNtf3yGwgU0+jyiqg1bartKzInDneTgm2pZGWI7ekbfrk1YTMT1MJaCIYpZUUdgsgJuJC\niy+vEcUgTXovsTVky0TaQ1sEDzSIHiHuKjY8wuHQYmKNXe23uv/v5UtqE1iawLV8zjUZUmuGbVau\nNDhoNzlozxMlHm/O694zjGcPYTlEo+J7LnkCNeJMJIseI5Fclb4umZcHFBGurF/gqe17QBaMY7qB\nISvypgoby0NG9YSz+1dxy3T+SrsEs8CiTIscRVlmDXmoaLoiIEg4LlgQxXXFe/pSKdVQiDKkZKNI\n19BueQYTA0WmjLSBGNGwKn2VIkTyEJhkUw7dhGneEFEcGbbrdmIcIoINyp03NlnWBcbD1Z4lzyxn\nhy/ijC4JmWV2esR0YAlOyEk3HxSwTWCj3mBNLOeqMwiCGE+gJSwnRJNGjAsN5AcVPT+iLyO6up5F\nlvGpMzs8MxoCcMZNuKPaxWjEmjkifQz7gOWsv+W4A7yS6aYKDy6m0eCi6XK3Ms80z2jliAR7T2TH\ngSyPj9E5e+mzv8k+x/qcBZaqWlUdq+pIVV33+err8fO1yC9Vlz/yCMunJxxdPeKh+AS/f9sf8ovf\n8I94z4t+k6Y8w7mP/EM2r+7RuIx3PvW3cWXgjtc/weuu/RFvue2baGZDXuSf4gcuzPipdz3Cj/zr\nD3Mwbz7/E5/oRCc60VeiHn8v9//h/8Wv7P0U5yYvpi4/jm2WvP3CO+n3Gt7yhrfw91/294/vdH8V\n6d+IyDrwk8DHgCeAX3tBV/QcSFXp21sS1MAIUSuMJKxyXzbwrsLEBmMbQMmaZCi3GrEdWCKakvPm\nKVBD5XLqrABddUlWGztPJi2qLW79WupyrR9ytRjz9OAiitJ2eHFPiZIBilvbpRcCElPxB5AjeDwi\ngrjmeKMbjKDOYowStaDVfdaOljzR9jg0W7QJ+EwmnpClFKZGApsEJX5UAAAgAElEQVTLA0Y7Cwbq\n2dp+AocnZAAOQyDTTbLeETHLE0I9PoZoGp/KJKGj0xatwdJDgWWcU557EtVAfsvDeGuS54x4c+Qr\nF6IYVAQT00Y3I2IlsG6WGJPAIphIo5br1Z+zCEfsjVvCMeYNTC8cB6FmKMQphV/SBKEpLItRQSws\ns6ioaCJCxjmVn1LLJq3ZQeOoew1K3XWFxBhUYWLuoM7S7Xk10E9zXEgIBNuQFw02KkW9wm8oPdPQ\n6y0xBGJM44w9FriioXDJpxY067xLHrVPky1mBAmp0G6P0vimjcxNL4FNYksZ0nWYyn1LXuwm71Z3\nDRhVWuvo1yWqEacAC9IA6pi9rOXKKALt8fUkSCLiieFa0XK5rLle1MTQJ2v7WHKKOOaoLTgMOwQB\nXz5J6LxSN2QbNZ6l6SWoBKC5kMUbBJviD4IolcB2XDCKVef/S7L0AEcrjvWjJcanv9mqagxwy5Uj\nzl19hq3DPa6PK3AL1sUTTeo2NeUB5+IlJjqAkDb80WYIafRyi+u06o5fq4gjmBXvM623MUs8sGjL\n7jhGTJfxlAePiuJtzdTNjwuSad5DBSwFoEwYEao+pkOk7A5G9KcBiVDa5E9yahAWmGL/+HpyXZ5Y\neq5IOU+IfRsB2zA88wxufERLYKFLohHy9ggbI2cmO4zMegIM9ooE0SFyeZQdQ25M1xkOziBdIPPh\n0HAlWycSiGaOz+YcT5MCzdBQLipW46Uuq1lkjr633RUuhLKPcd11JwryrDDwL7O+6n57fjF6LcLf\nulzw5kcKvveDI37wD87y5veeZnP/Yd5z4X/iz0//ES9/9ByZjzw8vo9HHvwWBqeX3PnND/Kmgz/k\nXdkFPuDvwO4+yD94yYI/+cwN3vQz7+MTTx++0C/tRCc60Ym+IF37yIf5zZ/5MO+9/o9pBbx5B1XP\n8p7zH+Lc4F7+zff9Nq8+++oXeplflFT1v1PVQ1V9K8l7dY+qftVDLmKM5MbRzpOnIIstGdCLwsb4\n5u8hNRCziBrFhBUyHMQlcEGmDajQiwnnDGmjdhhq9NjUblDXjbcZ5aCfjOImWspYYBVaLMEIDX0Q\nxWhkHJIfK2s9iFLKkLoLAjWihK4LlJDVCqYlDTsFEGinuzephnR+KiuIWI5c2tztLJLP6BiHnoxJ\nGA20IjBswKeOk2rReaigMYE9qfESubR2AaMZ18MuTUw5Sznd6J2Yjj4I0uGzWwIqXQqQEQppGYc2\nuVsk+cN8zIm+xyKs87RuIAtHG1b3z9MIWNXrtrR+iQ2JTGe1R8uA1uQEgay/ArsbrJ2jPYsvDYjD\n2R6VjAkqzDHs5ts0JgNVgipe+6mY7a4F0zriNLC2O+Nsb0a/bCm742aICZwRUmk90AqTeerDMQKU\nmwcopA5dXH1PTbGIiVQpgShHDGxDbgNLG4hDz9ykzLRcppSxpYwTcIGxSX6oVaFQ+IB3BWsI2/MC\np6siSqnoE7XokOd/0ZtS2QWxmOOdoTEZirJRbbPVDMnokUufZbgTZYRKuraUmkOzSetz9twaoRv8\nPDgYMDMeNQtW421BLEQDatjUCb7wBAxL6RGsJUYwMUMo2Z4M2GiGDJvutYlSELh4cAkne5jgydqW\nvfGI3QuOfjlh3L+fEJ+iZy4jQZkNLxy/7lWfFDgeZ6sANVtkVWREysjaLyx5m6AWjkjTXa92exc5\n+xRz05DPlwzS3QfUmHQtd37AAx1xI9yO00ARhZCwGLgODWN1gqIsyozKehZ5g4l6E1mvEYzFVi2D\nw5rtK3uISX4wl1e0KIu4ZLd0LCNkNk/h4z4VHI9vvozGCi1QOcfRIB2BUUzjkt6ByAQFivI0tXGE\nbNoRIBuqfkkmS4ymrLz0PtcVk8YSVYiDHYLNOciHPH7f3djTLUYjMUT6ewc8X/qcmPavFX3/f/Z2\nVANNc4PJ5ONcvfRvefyT7+HgMcvkwXUW7Ud54vRnuP3yj/LQmQvsP3oXjz/ccNtdH+Ce/H5+4AOB\nV9x/hbPXr1NnGd80HvOptRfxc/e/m9f+0Pfyg298efJ6nehEJzrRV6iuPXnEh996P0883FC7u7j/\n3Ds4fSi0G2PC0ad43eu/n5947Vdl1woReRXwtKpe7b7+IeA/AJ4UkX/8fPh9v5za2Njg5aM+kyuR\nc4t1XDRonu5095pdCjPHyKhDHwu5XZCFSJUPkeBA0/hRmmwSJMwRY1n1dM5dfYztwvP01unuDrHi\nmwLsjIUfEV1AgvL+zXvoz3Y5vZzixbOym9tuOKtkQe1SaCtOiIWBVkhEshWMG3LJsOpoRHF9TzGq\nyQ8Mi+EYmppoDdeLHTQ/oFUhj00qmnqWMuTMncc7iyHR8xox1K5hrV5D1IFaBmFANMlbFcRAGILA\nUbkNR0fkvkdt5mQ0ZOU+TgPLuCRKRrMwYC2aRyZ6wIgdAKIxzDdKturIimndiCGqpXUjWu8RNohu\nDiFQbSrtwlJrQZvShIhiaCiwWfLOSEJyYIxn/dxlsrKB9jTG1ISsgSrtLQbFaabhgKX3LHPH0Axo\nrSFH0qZXhG0ushknVAqtgCyEufTox4QSx0I/WESURgy9FlCPiWNi1sKinwKH8YTOm+cJbC3nLLMG\nLYUKqExLT2tUHEZhTRfk6w1aK6FYMjCBZ0xNPbyBMylra2EcC9OnDFMkwiZXodfD1w5o2ZxMWKz1\nWTSGpjfARKU1LTavsNZg1NHSQuiBWQCGzK8h4jGaipRWS0J5io225Jn+Pkc6YCvb46CoGC9bjOm8\nXCJMlxkPHtzKZu9mCdcj4LzwZMxAyoRONwFPzmExogg5SGTeK3A+w7dLntA7WOcKzy4FhUiuHmJk\nULccDZVRrPFZSWUdMUZisNS5ZbHZY21/iiEw6Y8odI5BKPWAQza4vn6aDbekp0uy9oin7K1c7Fqh\nPi5Rs338zJp7WgKD6Zw7yXhgFLk03qK3d4NBSNefKEhwSC2stQ6JJuVqhZpCIOKJkmMljRYaiSz7\nln6zAXqUut02wwL5smHmCiwei+KyinlZUroCg8X0b8MvHk59zDbiamjoU8kMFK6vFfS4gS63qO0G\nQsCY1Roq1vfWcHcfodmcbBHIfMZW9BTawzOl9oFI51aS89ShZFGUHJUWFi1HozMgULsea1VFNHMG\nN56/xshfiQKLh9+BTK9Q1FN26ik7Rwvu002u3voYT7x0lyeu7pD/iUU//js89D0/wm75SsrLwkPe\ncveL38+9b/gEnyrv5I+vv4LejQW31td46ZVP8+rHPkL82G/y7rO3sP0dr+fUm/4m2/fcg+2IRic6\n0YlO9ELr8NqCP33bYzz60WuonfGhi3/EIxuf4qWXvoXeuOWWxx7m237qf+TWzRe90Ev9UvRzJDIt\nIvI6EqH2x4GXk/IU/8MXbmlfurIsIzOCsREH5DjmJuKlopQ+dTxime3h/JhSBrjGY5iyWOxQjBqW\npgfSsvBnEGNoNNDvugajqmJtOSE3eeoYWIFYYQMEA0+sn2WNA6q4xGvgsL/NznKK1ZvBpyVCqZZe\naKk3KzIR1IIdTxnaNYpFJHS/Fl1eEnDk0220vEFGS2UN1/sjMAPWY82SguAMnynu4GK4ggkNOpyx\nX2QE9Qy9Y9pzLNUwVkdLRAyE2Yh+MMxHJf1JTsgMR8N9WrkDRcilhJU/TQEjZOoRmzJzRCpmYcz6\ntT7t5DTx9icBYaEVrZ0TF47QdVYSwDpig0LWsjAGpeWw1yOLQlTHNVNghoFbwu0c6o3UVVGlzjxG\nUv+u1CVzBtjxdbw2mF5g2h6wLefwMSd0CPaAouJYasoPUwFcwJgsPU7RY9dE1nTEtcUWi6ylT8Np\newblMpUv8cuCXAPIjMbkmLrARItXqKYD+vsZfuBwuqQuhrTimJicLFSEqOSizGkIVfLERZSQVSnW\neUW4E8Pc5aguyCYQNlIILO2YpelRhimbrSHPPOftgzzpNkCEQVsy1QIblNb0KUwAd8CiLnBrGYLD\nqzJz07SZJwPtsepzGVW8WKZFiZXU4Zlrj+F0zGJjQd6W9CVj3EbKqs/jJALnrNcnx9B2kQSKpz97\nEeNTu5TbByypMeTUWnRBuAnHP5w0LCplfL3GnAW6chlgZ3Gd/nzOtNxiZsfY2GdvsMMn9SW0VyNR\nFeOVmoLCeEZSIY2nXwzwmrpZpvVsmgmZdxRlhiMwsRsspyVxMKcMLUhkHBOtUVWIMSIqqCgxLLuY\nA0MlJffW65zLApmMyJhTXj+iOWOxEogYWlGGHHJoAd952aRKVULM6JUlR35KKxmZKsPFQdfl7VD9\nqkQ1rJ+KuP2S1qXzVcgaSEEwgeiGzGiIHRUQFA0DPnlxwOn9HsG0qLG0rgcIVe5AVr7DSD+2ZAxx\nAq1OGS9nQGR2uaS/3WPfnyIzgb3CsbHY56jcJGeOEBmoMA05BxvD5/jd+bPrr0SBdekn38+TNx6h\ndtAWhtm4YG/9xdwoX0kThAtPX+KbH7yfN370M7z7Nd/FB24J3PfeC2RxzCeqEffe+8fc88aHGH7o\ngA/c8i38XHgzjsDfnH2c1175JBcuXcL+4i+w/4u/wMPb2+zecw/yyldy5qX3cfGWWzh37hzO/ZU4\n1Cc60Ym+QrQ4avjw7z/Op997GaTl8TPv4N9dfDfbhy/lb1z9Ftbylns+9Sle/1//BNtf3cUVgH1W\nl+oHgLd0Y4JvFZGPv4Dres60du4Cj18dcKq9QaQgiiNi6LWep8sNzu1fxy0r2BH8MsNfKpmtOxb1\nmH3b4/zWHItNm68Ov97EmmJxmflaQ98PAKViD09J380RNbjeBFElBoc3CY0xe+Yc22cfo2VObjZS\nsHDesDFTZiKI80QMmW2RfoNiiBpBlcyWBIW42CJfDqgHf8oRA6ZSsiGWuRuAKg7wZcDT0N4xZzaC\n01LAUqmyEt8a1AkzFxH1KbBUhFJGHI4ucT7usa8NT28mr5r2xsTpNjLaJaNBxRCxGFVk7YBWUkdv\nbGao9jmsdjH7keyU59B4vCk4JfPk35Gb3TijMA/PULMFGKIaDIbYbFGwTuP2KKQiDw1mx8FwwUgs\ndr6kkjk9HE37IHawR0OLREElsrDXeWYROUM3nthlEWWxorA1y3KN3rLBuDR66HOYxj6lLsnq05gy\n4rEMsj5itpD5IUWMbOa3EUdHhLXLlPQxaokEFsuC9RiJBEw0aIxYY0nFnKAEDIEyr5j6QJAJTTYi\n+gI3nFJkDbTgJaDnZtSXR2Qmsj5reVXR49eG69i6RdWQxYzEn1HyeUOQjCwaesERmoA3DjUtzrRE\nnZKxhgdEHXV2A23H1Ai1MYn2DjRiabUgdB46VIkx+d02m1so/C4DPWQnDJgCbSg46vXZCS2Zs2RY\nokR6vsYMAi4sqTJLb9Awm0MaI/QYE7hw/QZu7rj7iYcYtIHdjRpvDAHDUT8n04Zz010eOzVATKTQ\nA9aWB1T5Nj1ZMo0T1tsjfLbOzNRcz9fYaI7INKP0LZlfcNelx3jo9D3Mmz5lXqEoA+kx8BuM6iPU\nLhg2cw57m93PdCASsSo0xgMGXxhqCvqmTx1KlDlmsEe5PkEndxLWHkf6C5pFRtmmIOYMfxyYPbU1\nmc/JTU1pAqwZwkTIgnK6epqYpetyNeKJQnKkBWLeQ6MltxdpZI3DvKF1GSFrUd+FpYsyEAMYlian\nNQGDUOVrTMvzlKYG6RPEUtg5jjXM2i6aDentVYiJaABfC9XBFs0oI/O7DJqaXjkiFd4GlTES5/jF\nkDVdZeZ9+fVXYtcf2gUbD3ycvzzE54s+RhRTLZlsbfMLb3gTF3ef5CP3vYpHT32Iv3ZjgxdX38oT\nyx1OveSd3PKay4wv/z/c+eFP8OjmGh8338w7Nv8jLryu4nR1wNd95kFe9elPcN/73gfvex9VUfDQ\n9jYf2Nri8NwZ2jtv59QrXsE33Pt13La+hjkZKzzRiU70HKtZev7sXU/x8Xc9jW8817feyx/c+k7W\nYuAVl97ERYWimvDqP/0Q29/1HWy/7KsGwf65ZEXEqaoHvg340Wf93dfE77lb7r6HP99/gKMHL7Hl\n0va+Z/o0JjCKnuYgMovnya626NJw4UiZ2oI2z8h7Y1qO6A0nDEe7xMOChj792GC3r5JlhxxdupW7\npjd4zxkhzlpyPOnQNVTWoKLs1xcpfIWGZzBRiLHFWElhuTtXqeQ8Nt8DWYAf4qSP2AS/cF5pb2zS\nXtgHI4gY1MY0irbWQA2H5ojc9ijqitItGdqS/OI3crD5ByzKEWv1mFiVYHKsBqwHCDgTaUMDm1cZ\nX9pA1p/G3ldx916PT8mAkdsi6hHqIuIszbAHByYRGU3gqcFF9soxt+9/jBAa0Jo2trT+kLEPDGTB\nIo5QJ0SFenwDqTOICVJdtYdEAZGAkUgksq/79MwYCWtk5oBRfp7+sMbETTwOP3wGv2epCNBOoRaW\nG5bgA3g4YEpgiOaKEwhCgnpUwnw7h8bhW890zZIthog2WAYowplwK+XkCMn3yDYnWAqqKjLKFrjb\nDghxALXlon2Sg3ZI7ZJXp3IFBUJmEkwiRiUzHlMJcwNYYby1h328h7bKYX8A+RHF5uP4OCADCrtH\nIznat7BwXVECd08tD/mAjzleDTkte9UAgCoWxDxVSqncy5i4EdZAHrsulUQa0+CdUIUp/bbEOSAa\nvGT49gKtnSFGCAS2DyYMru/TH1h2nhYeH5/CxsdpnaEZDECFqtcwXR/Qm646MR7pZdhu5PDPhrcT\njaHqWwatITdz7n3mIa7p7TRqcDJG2CefXGBuLqEamI1ysrni1CMYnIE+IzaKy3g1ZG6d2OYIDSqW\nG34HTE1W9Ml9TVYviSib0wNQYZBVWCnwEukTMNEzVGGukVxrCMnj6E0g9RX/X/beLGa27Lrv++3p\nTDV/053v7YndbKqbFEMakkUxMeJJEQIkgoUoRhBDBpwgCII4DwksJUagBxvJUwLBBgJnABIhgGEg\n0UOcGJYD27GFyJRJilGT3Ww22XfoO333m2o+49575eFU324O5gBSpIZeF+vuvU+dOnWqvqp9zn+v\ntf5/DyLUSYI20NJyvT3ijXxL2Brc3kN0DOhNr2dlrKeTDhsVWtagcrxRpE2HDZFCl0SVIRIouhmF\nPaUNBSoKnVJ0OjKwLShorSK6hpIhI1JEtWhlKHUvNPCuyG+XRmJn6LDcnV1nxDldn+yKIhKVIrSK\nZirouoDhkleb15iXrzAfanSyxl0+wcwbdGuZieHEOZJ6Q5MqmvlV2r1HQM/heZIcckUeI6Kwuv6h\nzdl/KC4838mu/Td/jeP/9udQqkPChrh9QnfnC6iHXyJWfcHb5PyMX/y7/zv3nx3zl599id967mWu\nv/G/MJ3d4FL9M4y3H+bJs/8XB8/9Fh/717bMvvwS1+/+Fq17xMwueCZc4+X9a8w++SdZliPK5Tlh\ncZeD8ztcf/gavPYaAFEpTkYjvjrdYz47YDHbpzvY49bNa/zUR3+c2csvYCaTH+XH9YF9YB/YH0Dr\nmsBr//g+n/uNO/hKOJ79Dv/45b+HMSs+efIT7Ld9HckLt7/Kj3/+Cyz/xE/z6n/2V37EZ/0Ds78N\n/BOl1Bk9VftvAiilXgCWP8oT+0HZlStX+Jd/8k/xhQf/HBrw7YhMNSQq5VIZmDcvYYsaXzsyU2O0\nRUsBCCIGJRpRfWH4aHTO6KHBdhtG6ZYyRJQv2TM1QRRGpWjZEMhwCogppclpTYKPCba7jd9p3Xil\nuTsa80JX0zyJBHeBaXNERaKEnsIc2IYnTMsXmPMO4570GW88j7qXaDQYbeiINNMLBqWQuptUXcv2\nxhnnxYSDdsZKGXThIDY96GtLMj2GAKIFHztcW3GreY4z+xbpQcPJ5I9hkw7BYFxLuryMXQgqnICO\nNGHFeTok+BU5Z8zdDBMeUrsJVbohTQfoSpBCEVxGIp7tZEU2bfGP9pDVGWem5ry9i3iPs3DOkEY5\ndHzAVX0NUZDqnGBKtCiUCUzkiNPJGbLYp2s0EhUFNe0wwW1a1m7AKD1l1U1ImhZrEgp3SJKMOFEn\noBR+lIF2pG5MtIqs83gLCQnBj6j37qB0n5sZvVDopO8nnlGcIO2cNAbKJwMOJLK1GS6M6fwKkoKR\nHaFo6KhRymNri09AjCd0GYPTJ+RHAZtsKL1BDOTNgGgtk1HLaVkwjS3nG9OnePmayTqyjopRBB0j\nYiyDrSfalFYqFJpAQ23GoEMPBiTB0dHoBtflFIuSovMYSalsJAkWI2NauyUajZeeXCUAw+Nzkskx\nn9p8mei2HF++yXZ7RBrOEZWhbESrJWk0GK+o0wwlAiGCLwiF58CtmHYNk8cNxF44WPfCcyR0+OYq\n8+w+djihtVsGnScctNTJpf7d2JqEDoPwZHaF58/usm5S7s72KSuLDQkrp6h1QHmLCg3KelCR1HlU\ntFg02/WERKDWCa/Mv4TUQgwDpntrEplT5DW1PcaGCXZUo4cZiKfWBftqQpknjPXbqCAkA08QR6NW\nCPs8HmfouesJXZRiqR6ziSkqb5h0WZ+q13Zk65I8bVH0hDs1Hhk9xKiMdWYwi4QyejolkFioAl7V\nBITXr18j0lIrhzZ9FHhrRtRuQVqsyFRGoyLV0BDXltPxgBhhaxeIhoEsWakBh13N2lZUOiEn0iJs\nCotUDeCIdS9iLvTEGlYXpNEzKStC/kEE6wdqoVzTXV4TVhs4vsBuFPbZP0P68i8A0DZzHq0+S3Xx\nRcanD/nF//PX+Zv/1i/i3CE3P/f/weiY4+mPsb/6OYZPfpr1q7/G8x97jUvPz7j/zqucn/4Eb4SK\nu2bFbPQa1yYX3Li8ZRASBs1lfPVpqu1lunpLs32I2TzkYH7C9QfvPKV9BTg2lrvDGe3wAD84RGd7\n6NkB2c2b5K+8yPQjNxgdDCgmCVp/EP36wD6wP8pWVRXn5+e8de9r/M5n76C/dkjqC+5PX+e3n/97\n7KkNnzp7laS7hALGpeLTn/kHDM/OOfnUT/DpX/2bf2jIeUTkryul/iG95tU/EHlXiQdNX4v1h8Ju\nHT3P78YXWJ4+IuQ3iMUCJZFUgck+hFdfwiYdzjs8ARkfUGznmHqBUkLeBT568TpiniGvI9W2Ra4G\nIoqlXzB1p1xZBsqjq0hwODrelXlttiPEAVHT0VKFAWQltWQQF5yHLRu/BioSMrJW01gHUbFUcxLf\nkMkJ8vgqdSYEHTm1F3Ra0DGlGGxITUORCIw0Mi/QxRnztEJJQqM0WhmUEWoCEhO645eJV7ZEEwlS\nsbEt09FXODGaeZeyyvaZjDqqaNBRoZuMPER+++CQWw8OGfh7eKeIbDDW99GT9Yazgwbrl/i8obYT\nnJtAkuLXGRycUSkwyZb16C3uxCFnVzOK8xOisSiZMKwOUS5lKx3ROpIup7JzTCHYbQCl6MhwxYo4\n38dHIUsbgkClhdFMw1pYmAziPrNRikTDoJrikwZivzCs3Y5FMG+ZjhSla9GiiD4limWtRjtCCPoI\nRezrT8R5Cg9GXcW0BdpPSDNHjIHz8zGPPqS51uyo+9WMGG6TRksSBLWu8aEhdjVbvc9+co4579hM\nNVWicbsSN42jVRkicOEdTYjcWjSYdJ86lujYi/VmxmMGt3oAuPkiiysTtN5gJcOtzpFwwt7JS9SH\njnF5xHY4x/me2CBrPHlVURUZacwwcb9n/cs68lXA64ZazZhu7qJnZS/RbD3BtcSoSHxGWXXs+zOG\n0tAke3QqZX+4otsJCyTdGEuFESFrWwQweCa+xImQkfLh8ze5c2S4SAJaCZ3qqLMlba7JVMmy2EDT\n16cZG5EYOBkfcGZ6PsdZBXG0pRxn2Mqh3ZaL6QEivVSaif1zY0+ojy8LRn7J/e4Surb81JPbdNOv\nUbmU7qolU2uuq8BdNyF08PoBPD83pMmagQXfCU07Rud36OwMwSLSci4Fio7TXHGW1tR5x0hZhiLE\nqDDSCzonvo9oKyK0mi50nJklSxmT7085NSmpLxk2Gqscte7TDlvd4LVCByGohqwNBBpIUkojGAR0\nw8neAVs3JMtnZMUFWdvT/l9yZxwuH7EfKn5z+iqL6RpOFJtpyqN0hQ4DLkkKypDEnr1QKQ0Io7Ii\nPbvg7NrB7/1EvbM/EgCr/PxvUv+N/wKA88mIf/STn+Lv/olDivglfvzslJfXR7y4/Zd49tK/SvaS\n5Yaf8w/Pz/kb/85f4hN/679k7+GXufXkK/CVX2c1ukV2+y8yfPUM/dL/xssf+af4uuD00U1OTq9y\nv/ow97F8RtF/ujayP1hxfe91npWGa2hE36DRL1Prgk03pCkj3XpLWF9gVqeki1PGx1/FBP/0PQiK\nRbHHcnBELA4IxT7dYI9utE83OyCMC9TQ4iYp+d6AYjQiGRaYzKE0u9UCCL4mhhYJLcE3hNAQuobo\nG3zb0DUVXV3T1hWh63r3nuA7ovfEGNDWYazFPG0tNk1xadZ71nuSZV+/7X2PWZf8obm5+8C+O/Pe\n07YtbdsLn34rB9Baf9f+g7TgI10T8G3YtRHfRUIX+tZHQvfutn7frgm0ladtAt2ubSuPb/ubKK0V\nSvetNqr/LWrVCz3qgNaCNqBVxGhBqdhvo29FOrqwpGtXNN2SbTfnkTzia+Yx9/we1+Y/zocXL5HH\nWzyYvMndo3/Ccz7y6bOPocmIojho9vjUm79N/ubnqRLLyc//G/wrv/LX0X/IyHhE5DPfYttbP4pz\n+b0ypzXT/Rdoy3OWJiKid+u0BkeOVq+QcopUS5Yzw5V8n8wfcbu7jdGaUWgxOGy7JRl2nO1HapMi\n7ZZyUNJMLhiFLXubQN4oHs6uIaJ44c0LHlARDh2lCqxtSlPWjNMC99jCjY7zcEI6jqioicqjVEkq\nQ1oldKWju2dZp5FAzqlEdPIO21ZhlIAW5PAeXavZ+CkTUmy+wORLTsThZExQgg8a6TLqZAXrESaM\naS+m+GxLlZxR2IzNjTkpQmIuWCcFtTKIpFzanrNoL9ElK8p6DSYAACAASURBVCod+eLVI659KWf/\nZgMmEsXw5NGY40GLswqvDVUrFES2Lsd1fdLkk7jhMAwI5oRoIl+5ESmOG0bXKrJauJgfkYQRjfE0\nYolGYeI+On+EtznNsGCrPVmVgo8E17GdCs72v8etiuwVHqdLurN99HjFG7nn5nnOSMYsij1uMOQ+\n97AD8FHhDx9jg8EXFak4kns/RjCRRiKJmiCUpHJIYvdpVU2qOoLOCDFBp56ZvUynNhy6HCvw5fI6\n1/Sqpw4XQ2gv45THdQFvAtODU87vD4l6r48SlA5VGLzReNXDgPGqYaZmlGrC29WzDJoTOjTZZo0K\nHU0ywpkSgqFIHrK9PsK+vaBVCVEZ0lizmVieObesQsnSHBBVn0S2FcsUaGLCyBvMxuCzgPYF0JDV\nDcS3UKoBFO3oPpZBLyathDKJqEYQrWmItE5RNYamSzFZSZLWtG1PrGCVYaAzotJMyk3Pjjc6oths\nGXU1QWVE0SDgkyWd0hgR/HyPZ7qHfHV2ncFOmFgpzZP9CVfutfg0oqNCtELFQNHeY9QckJSKS8uH\noISPP/oSX3j5wxAUg7mmtgEjggTDWTVDZwUmGOx6xjx3bJMtCGR6wFACHynf5n5xHasjo8Wc//cZ\njR1d5/nzUw7VQ74w+QgNCj+AVQfzLKVJAhf1XbzRZOmCSgYge1ivQG2I1oMkaFeTdBHE4tj09aBh\nSjCa/b37zE+uMDBDvPK0aCINstO7mm4rzvJhXwtq+kgdOkJkB9os59llrmdbVPQ8V7+FZg+XbpEO\nEE3mO1SScn+6x/F2n7xxPB+uguoIOlJfjGEMEUcaDV4GPEqvcM98ALB+oHYyg6/+7KfopjndMOFF\nG/lPH/wjlAoo5ZklX+WaF/bimmG8j0vv8Ldu3+BnP/7f8Vf+vb/M//T6/8x+OaWZjzAXBX7xBu3/\nvcZ+7s/hXonIC5/BPvM6V557E8oR5f1Djt8pWD4ZYExCXRR8cXTE71qHCp7p/AmXTr/G1dOHDJqG\nPGpG3qCDRgcgKkSDiIZdhEshUJ4j1Ryl38KicKqn3SQEcBl6cISZ3sIcvISePYPOprTaU1JSx4om\nVnhf0XU13lfUfksb617ETb7eG2lpxRNVRBuLNnYHpkwfKo4BkUD0HcF3+LYl7nRXAAKaRqfUJqXW\nOY0ZUOsBjSmoTU6jE4K1YC3KObAOZR3GWZLEkSaONE12npJlCUViyRNDnhgKZygSS/a0328fpJZB\nahkmlkFqsOYPHuX015lIT8ErEbTdFVd//xZjeAqeRQStDdpotDFobXbilfIUEDVN87R9f/97eSy8\n7/vxg7JvBFxJkpBl2VMfjUaMR2NSN8CpHBdzYm3ZLlu2i4btvKZcNNTLFvERo8AARkVyXfauSjJd\nkeqaRHUYFUkIDFXEKNm52i0YGHpiaANolBjk6dggYumnXYNgULS9qwZFg1ErInNWrLivt9xRFffY\n8CQ2VGKxzZC97QE3qms8HxN8eo/N4DZkhiwZ8NL6ZZBI3qyZLM8ZnJyitkvelEj14nUGr36Myy88\nz2d+/e+QFgXZcMRo/5DxwSHD/QOsc08/WxGh7iKbxvdee9ZNx6b2/MmXL2E+iKL/0E0rxUf/2Cd4\n++FnGa0fMJ8fEa8+wncF/XpDwcTf5IG9w2tHH+bjYtm3Da5T6G2Dyw21gdQ3BCPE0KBig7Ibxrqh\nySrGyYJ0E4AJLnR04lCxIg0ts7nQTA9ZDScM1xbjI/XZMZPhKTHJWbuMIBcUSlNtDjBZw+bsear6\njEJ1BNFUScTFwLlo1kw59KcQewKHYbJm3Q0QI1wd1ngViTHBBI0QCIsbqLpglZ2Q1XtkQGyGbCuB\nS3Oi1TTKoqWlzBJOsis47Wl9yqRtWC5z8v0zbHJC0Bu21464c/WAI12yxfL2gcFWGmU1ttUYIu/k\nZ6R+SH3vozw/e0AlHXQJISZUe3BJRWqXg4a6MBRlQ11OaUjwmeHcn3JFX6JqbzCPJ2hJsbGlTYW6\n88wHd9DtAOEYFQtiseEk86SlwzSG6laCnLUoleC1sD9b41eOle1QR1uuhhHWz1ChJIuOGCfYw8fU\nZzdY+5KZS9GkZGYfkT61zUuGOIMNHo3uZy39NqXaIwdiV+CHJZ3PaE8vM6ngwsyx0wekrqOKYDKP\nbRqUtCgCVzcbTgcTRtteUFihmS0DjXuGrRWGEkgmGwq3oF5com6GtG3AOjBeuJgVFJlFRLEpDIdB\nM7YDlKoYd2eobM3D+hJzIreUpuvGWLFYX9OqY7b2RTIDpilRKErbYgJYXdNaIDYosZgOLtwTBvmY\nos0o4hY/KXHvtKhbpzTjCdtWeG32caJSWAKvrN+BsmbjxqRL4Rn/u0zqa9RoIh2iDsjKFePE8yRY\nJuWUIFPM+pxXQsntbIDCgdHQKr5y8xka3aeqqQimPMW3+9zYKIJagkoRVSJNyifOLvjieMSVJ084\nTFfczS6TNx6dvsJRvaRSnk17xPb0iPL626iRoDZ3GGY32QsdV3D4CG/MFDbWRL/gK5M9FqljkQ0o\nnSJJlqhFwu0XIs9Wp3BfyF1Na0tmmac+HqJNRaEWrAcJ+dphzZqNshTJipQOwvNge4CUSSSMT8nS\nx5xOBqRPUk6LYgdoHdSvc71+gYd5ioScNgqF2nBze4/Hk+fQ3XWiaTDmlLpXiiOVLRIHFNH3tPk6\nIyNirjZ098a8UF3CWqFhJwi+KojtikwXdDiCpPzu9WcphwUi8kNZ4P8jAbBi+mFE7pJd1AwvGkZm\nyb4558C9w9S+jVG9wFmQjEV4kfvtn+P+/MP8fLnlf/3UTX72lb/Kv/47C14yS+yhgpml8gNSteTW\n4/8Dd/+czvwCo2cFfe0rqBe+xHMvdai2QB8/T/NOyuLBmnMim8ywHB4y37vGmy8K6WbJYP6IfHkH\nZEtnBeOERAspkVyEQkVGEhm1wqATMg90Ct9ofK0JtcE3njjfEud36O78P0/fu84gzzWDwmLzFJ3l\n2DzD5AVmOERlBTGxeKXxpARJ8CQELFEyICeEgoUMOfEZJ2QssKwQVghrJWw1bJSwRdjQt9+ujFAJ\n5AJ2dyuqw853T4qKHvRRElVJpM+l9qrnHfomtpJvY1YgRZEAqdJkWpFpTW41hTUMXA/KhplllDtG\nuWM8cIyHCdNhwnScMh2nzCYpRWL/hamZMQaa7ZZ6u6HZbKi3G+rNmmY1p15d0KwX1KsFbbmmq0p8\nXdG1DW3T0nUB7wNIH3J/V+BS7diQtJLedzKEuk/9RqH7SUKpPnUgKrxooqiepSuq9/pPXe0YodTu\nIP2Y93k/1sh7L9R/6E+fp3aMtPrpc7XSKK3RWmG0Qu+iNw7IEawILkZsjNgQMBLR/TtAI7gQcSFi\nQtwJGvZF1urd1wO0TlEmQ+sMZRKUpIADlaBwaFzfKotSZucNWl2g1bK/kVC9ds5EGQwGvVtZZNAD\noK+3FJh9x++YCDTR71icAlEi8u6ixbv/JNCP+jYQiNLuzmmIYYIjIdcFmR7glOK5AM/tXsPHjrN4\nzkN1xul4w8VetRNf7U8g7zqKzZZiuyFZnuKrGh8DpYKuSIlOo41jc/s2j95+m34Z8JutcQO2yZiV\nGTLXQxZmyNYMaHVCqxPOkn1EaV77lT/DOHPf8hh/UE0p9TPAr9J/Ef5HEfmvv+HxFPg14BP0Kki/\nICJ3f9jnuXfzOnf0i2TZG1yn5cEXay6uDqmORhRa4cuS+0cfIl9YaBTlBGbhjOzsGtXVE7RV+GTN\no1jTWQMqMOg2TA4ixIq8NAzSinDhkD2Y1HPOmglpbKgzCCbl7pUrXD4/J7/3CIkt6wcDkhtbEtli\nK8XbtwuuZZfABHCRi2RAJ8KApBfblYaNfUTcXoGQgsB+ndClS4bugkznxDCkDPSRjHqfqQE2Vzm2\nwlJ/mefCAZUTYrellXuE9QS1v2LYjKmUowhHrPVzWOnQpuE8tgwePyDf/yr4SFU5lqMrKNNwmnge\nVRPqw48yOXmHaTVnbAydb/l8e5XoNa/o2K93iaJTkbZec6Fz0mAYTnotqhphdLRmffoRQl4Q/Jbz\ngcGP7hFp6bThbum4sb6B27/LWWwwRcpgvMbYQNU5HBkL1XDZ7XNwOXJaLzlOI5hpH+3wY8pBy0sH\nCaIUxkKtloxrS7o+JlNQqYLtlXdgMyJGYRuvkj8eYrxHXrwLSYtsL0G7pVtNiMUxSgxVEpDWcSBX\niJ3Fn+/hOgttzhEDuv1zstZQ5WBnGn1WksiWUDnKswRswRM/Ye/6CUZHZlcbQrzLHh1KIta3tNbC\njQ2jbUmwhs5WPNg/5FiOuTbOKa0QjcLrvr6vHTqSTcuzX9tSuy9wK7HQaA7K29SDqyhXcXB8jN+7\nTC77rJUACySbohpHyC9Q2rP2Bh0si5OGay8HstEJdgj3z24i60e9BhYL9Npxe/gM3gRMm3JUHpMY\niEZx5dYJbgUZSwbdG5ydvszyQqPTDaO9AaP1LW7tP2GdGKo6Jw8eX75DKG/RhWdYdo7T4ZzDYYqp\nhEBkfPGQI8Yo8p4Rzw/o7ENUm6O4zLBa8FPlfeaLFrt9zHis8Canzg8Z6pxEaholnIaCgjFhGDBR\n8cRUlF0kARo55yzNcekS7UpiTFgWKVtWOOt4aXmX28NLzNUIMdfJi9ukezUTFZkAzZUv026FSYi4\n0z0a7ViEwPpgxT4dsi7p0oTV+T6XZgsae04zgbLRmHzNqRbuDwcQLa+eHvOgWvOhbcujW32N1SQ2\nfGh5n0tpwqBa8MBdYjTcoJTmlc0buGZAos4pSOl01s8hsa+r/Er6KtfMA/LsOk1rSK1GV1seTlLq\nM0UzMhTtJd70n4Z8DxvcDw1gqfdS1X//2ic/+Un53Oc+930dI/7KEZrmvbHkeLlGFZ5jzbPU/kNU\n4Xm8GDzQNCVV6PjyLOW//8l9zkaWg5XnI08uuHm+Jl8mxE1Ki8KrljR9wDh5C9NsUO1Vrl0Tjq69\nhTn6IuJKiJps9QzZ+hmoxqy7irmbU+VrdFqSpCVZsiVJK5R+72/StQl1l7HtMi6qAWdNymljOasz\nYrCkQZFFIY+RadOyV3bsbWtm24bptmFQ1mRVhykjVDv1wW8wnURcHrBF6Ns8vtcvAjYPmESeBk9K\nlXJupjufcGZnnNspJ26fU7vHudnnwuwx1zMUCalAGoUi9MAqi5BESKLgdm0iiiSyk41UmNhH7WwE\nKworgo6gpWeuqREahEagQagFWoRG+m01QitCGyMd0MVIF6GL76reg0foRPDfI2Cz9BTCVgQrcefh\nqbvY4aTDxhYXWtJQkceSYVgxCCsyvyYPJWl8D4YG1bNRRdX7UxCEeqrArkTtssK/dxPkKTBV737G\nSqOkBx2GHszsYAdqV+/Qr20atNI74KJRqgcmPYDRKG13++7ynZUmaNNfJJXQ6fgUrCll+9dTBqMs\ndveKFoMTjcPgcDgsTtm+rxyJSr7jewziaaWlkw4vXQ9qxBOiJ0rXt7Hrt+386T7iieIJ7xsHCQQV\nCRL6x56Cox0lruovkEJ8n8bkrvN0XpX+pkxFREL/d9j9H3TP3CXKgrKgc4waYGVMnsxwdoByhjZp\n2boNre5BkRPNYZxwOU45kglHcULyLdbKogRaaWnUBqdPGOkHpOoOqfoaD6Xl9e4WX+ue4V53hWVM\nyFTJgA3Dbk3erEibJSp+fdTx8n/8q4yHOT/9wj6zQfo9fw/fb0qpz4vIJ7+vg/yATCllgLeAPw08\nAD4L/HkReeN9+/yHwEdF5D9QSv3bwM+JyC98u+P+IK5d38ou/vavEy5qNu1XeNhOme/tc64+Q5nu\ncXB6wPEq4WJSUrsR4zpw62RBTkOZjoiXHqCTEx40D6h1wpPzAc9OLjBG4zpPQcMhhuPjSwQs680Q\nUzmiS2iBxzdvsJWCThzX5sfcOLlDpiLZ5ZJYNDTne5z4Q1TIOXQjHvg7bA/3SOoSMzgi0Y5lXLOO\nj7Hmw8QsZzx9gDUG67f484Tp5BTRGms9utpHrW8RpGNVH0B6hhu/gV/PSNqbPLqak3Cf/e4Rd8bP\nkhc1adwVsnuLQnjx5C7rxR4yWpCFlthVPBxeoVWR6lpDox0XZYpXHUjOkY+QWk66gse7NZ5nTxZc\n3QQOeUzqC8Z757RAoGOGwytYJpHWzlBxQtdZbKzJtcW6hk7VBDE8PLnBfreHuBPc7G2sCGKEfV9z\nvk1I9xYs7R6JsiTAvfUeVrfcbC9RSGQwnKMQYrR0umGXncZk1UH9GB8TVlLg7QhRgaa+yUReQkKL\nqUrUWLFI7xFUw6gsGFLStXMeJw1F+zybLEMxxidzXlhPCZsIQcAaxsM1zfiM1USo0gV+/RjrG5KT\nQ6rNjJWr8SIMDrakhcHqhKBDH1GIgo4dPrOIEobrllIr2j0hTgpsKyT1Bt9qmnqPLJTMsoLliSZf\nlzDa4vYbVHBgFKYTtnlOVpdsHk+p5BqD9pDzcUVQJ2hasJa9eoFWwrKYks02dGLZpEPmxYSjzZrM\nN7i4YNrM6bSjjQVf3n+Z2CnaOuGFk5YjCUyO3mbYnhNDgjI1tSlo8pS2znveQ5cTrEYpEPE0yzHN\nYkZIlz1jplKcG8vbh5dQAoW3NLrkEydfYzKFan4NGy3gCYnCVTC1G/Rwy3pZcF5lLPKapNZMzZjO\n9anlWgvEyJN4ynh2CnnDVHW9jIM4vjx+iU73+m2YCuTdq1BKV47YrxbcNF/htfGL2NYSrEK7DR9e\nvkGBoVWRup4gSpEGQ5pUnBxfIk1LjK3xSvNoeJOv8iLjEDmYrpg7T+U2DLu+bm/YrtjonAz42Ooe\nDxcZH9u8ylJrfufyhJfXbzGLS9TlJ2wl453By6x0xgvb26y7OYdPzphOl6By5n5KngllOuTN6Yus\nRBgvDoleY/ryK6rQclFWJNYzWzWQfxSUw9uS/Tn88l/986jvIwPju712/ZEBWGd/7ZfoNs+i1Igo\nYyJjApEzu+DYrFmGnG035UErnIQ+MlNqoXGKOlM8uZoxv5YTht+wahsieEF56cXZfMR4wfhA2jbk\n9YYb5pjnBo95Zniby/kd9tyTHa8SqC5Ht0PaZsC6TVi1CcE78pgwEEhNiXZrVL5BTxYo915dVrva\np754hvriecr5czSLG0j45htRIRJUIKq+hYCSiIo95acNARtakq4h7RqSrsaEFhMbTOjQsUXHDsEj\nJoD2GNuQmorcbRmaNWOzJDdbnG0wNqJdRFthZQrOzYQTM+WJmvFE73EiU05kxpOdnzLBfw/BVIH3\nNP3UU9TwLcbqXWzSbzcKtOrL3nXfV/rd7fT997t5f5/3IjfvHuPdaI/0Z9Wvbr57ktIrpgMmCiaC\nDYKLggs7UBl60FkEYRAiRRCGITLYjYcx9m0QihDIg6fwnjQGjNBHf6Qnl9WA2qV/KNWnZ6DNDvCY\nnklKmR7k6G+uvZE+5kJHwKuAJ+B347CDEaKEuJuaQfV1U/QpjAow0hfBOiwJPThKVA+XtPr2qZpR\nAl1s6GJDu2vf79+8raaL74GpIB5R0gM/egCnd+50SqJTnHq3TUhUitMOrRIUPcARsURliLh+5VSE\nFkVNoNa9Zk1p1pR2wdbN2bgLSntGaUpqU9MKKApsnGJlTBLHZHFM4ccMmxmjesqgneF2OhyiAt6U\nBFsS7BYxa4It6WzXf293ljYNexcXHD454fLxMZPlshd2dAPq0ZTteJ9yukc9mtEOp8R83INSsSQ4\nBpIylIyRZLjd7ywSWagtZ2rOhZ4zV3Mu1JZOKVocjVgasXgvKN/RBsVb3T53imdAKf6Hv/BJ/vRH\nLn23P9lvab/PANYfB35FRP7sbvzLACLyX71vn9/Y7fPPlFIWOAYO5dtcRH+vAJZ0HZvf/gLaJNhr\nz5FcH9FVa37j9b/P/dMXCecV1faCrw1GdPGUo2DRdc1y/ZBZu+UF93neGQTeVIplMuBj6ZZJzNh7\nUjOkJNkIsbGIVjzpZvgqR0Vh7IT5MCWayN3JDVqf4vAY7TCqweiemawzGWd2ikiDcoLa1UvqoEgl\nQrkiqEAqQ4gJTiyowMpUAEQNiMImAUeC80IAot2gjN/NQBrdpRgdCbrFG7Wj8wYV90iVI7aWqxf/\nnEQiYvu50rWeEAzLYsC98TUm/h0aAmHgWGaHjEiBnFz6BVkBFqZ6WiMqEdLNqNdZCobRcIMWg2/2\n0OmGmPQ1pghoHRFRuLyjjZouRM62B1ilkaiZDk8YKAEc7XoEleAnJ5g0oKzb6Tmpfg6XngTABEsm\nKUbFXY1XQ61r6vkeo22D8gbGNTqRPpWq2yfmKbZLmUdDHqBVkEzfIfMtozYSlefE5FRes+dH5DEh\no2CpjymWmqxQZD6jTFI6bcnHjzBErFQE19Je3CTGHBM7qnRN4kENa9JYoYikPhIVLIsBRiLSphjT\nYEOgNCMyu0UDnRjSGNGmAy24GOi8w1Udynr6cj2NFk2X9dcd5RXKQ3+17e/PmiTg0YjqKcyJvRBv\nr7IktM7RxYwEECxJbHv6CB2QraXpUkYbxQuLhxzUa3yTkD/3mKcCUTbgraZxhoDFBPt0zvadJegK\ndiyJRgmNCU8zPqLWrJKCPARM9GShIUQNap8kaESnRBVxBPKmZs6C83vPsEkFdOjrwrxlmwVmLiGV\nhPNmQxtKiommCIbaJ0RalGyYHJ3wxt6PEbRnaE6Yru9TbvY5nt5k/0FgUK+YHZ3x1v6HUH6MN4KN\nDqvW+NKi6PApBBRN4hlGjW5tT8UfNSKaU6Ys45TERAauIaOkloRJ/oS86timEeeFDy0aZu0TMqMo\npGDoRqiuBtugdMlGAiItUWcIDVW3pt4OmDTL/rul9vlq8jJTdUbtIvtc4f7hFd7FSlv3Ts9yGXNU\n0HRRc6k8Y+3/OGiN6JaPzW/zF37pP4fvo4b7A4D1Leyf/tpn+aXb/4xld4SIQaIhiKVV37rYewjM\n0OyhmKGYopChI+SBkHb4xBOcJ9hIbQwrk7PSBRubUTlNmWm8+dYo2YSAlggoXIQsKsYepnXDuFzi\n6gX4DS4GXAiMvWUWHHu6ZZYtGOTH5MN7mMldJO0ZdYgat7lGuryJXV/DrK/C5hAfHK0KtCrSKOkj\nPcT+BnIX8fECIRhi1MRgUMGiQoKRBP09ZpLq0GB2bkODDu37+g1WGkysMdL0II4aIy1aul4wkp1o\nJL6PDyjpb3iNo04yqqSgSgZs0wHrZMA6GbJKhiySMaXN6dB4DN3Og9K94rxInyaHEIT3xiK7NDrh\ne/05KNWTFmitUOa9VmmFMhpleqAWjSIYhbd929keuMnXgTm+YayQ3fMx76XKKRGSXUqdCwEX+9aG\nvrbBxL5vYyD1gVEXGfjIwAuDAIWPFJ0w8D2oy4Mij4o8alIxZGJJxe5WUJPvAhzFbwJE3wiGQmiI\noUJ8Cb6CUEKo0b5EhxIXWxIEh/QRNGN6gGj7+jwxDkyCGNv3tSUa16/SaU1Qhl7JxNDfZmlA4bUi\nKtNruhhNVLqnodX9NrRFKYs34DV0OuJ1pNOBWnm6KPgQCG0keN0LJAaLDik65NhYYGSIkQF9BtnX\nmxDR8QwTLkCv8LakTTx1rmmy935XKkaGmw3j1YrxasVotWa8WjFcr0EpGmcp05RN6linlnVqKBNH\n1AqNo0iukBQ3McUtFDlxs0Bt55iwwsQNOm7RVGQuoUimFNkRg+wSw/QSzvSAz0vHOp6xklMWnLNS\nC9qspLGaQnfc/Dd/mcYMCDHy8ZszLo2/P7rb32cA6+eBnxGRv7Qb/7vAT4jIf/S+fb602+fBbvz2\nbp+zbzjWv89Oh+vmzZufuHfv3g/pXbxndRc4XTdoBR7QsWbTPMSQEMqak/UWf77C6JqoNdKd4Ts4\nbyvqzZohgaFUbFSGT4dkDEhTh6pKurKhSnoq7HWM+LSgDEIRPUYMjd8S8jHNIKPpAqqdYJKMvI2c\nLufsT3KSPFJVQLR0maVdXvDiyJLnE9r1MRfbSJ5qimyK0YInUi8X+KDRSQCVUzVr/CYwqIR15jjV\nlr3NmlzA5DkdwihsENNBW6KnE0IoyTwMhjOuXr3KtgksNzWlBBRCrgNp1+AZ0GqFj/cYUtHYlC2G\ni7WjRpC2J4gYqwHBahZJQdcKhQ64JCXRAbEpLq9paCnrjlJKEqcIneCdJWaW/W4BRmG3BUoHEm+o\nzBapHKVYWrdmZBOyLkcST+eH+E4QlZM4wcYSlRbUEdoALiuwKqFqAna7JcxG2NiivWWgcmqBpCqR\nEBjSInFOLCweTximOA15mhGcA1KaCN4rpK5JiwKltvj1HLp9TFjiCJDeIFGOta8pw5YY1uQuoHRO\ns14T0wDSoUJH0q0RMZxGg4k5ickYpIKxNZJ1rKQj8xNyVZPotqfPUwlGDCwX+G5GbiKCp40lyjmU\nttiwpiXBBwgGRFlMU1NNpogaYH2gVoYqz2jNGEPAZH16XeYCm+hYVJFx17DvNuR+xbXC4uuMummI\nqzWc1ZT5gC7NcU6xMjmVJMx0S9uBmjfozNJFi9dTxpyRUoGKhExoQkcSA7WMMFrRJgnaCVauoL2i\nTtdILYzcQ1SyIWjLsNinjgVzFHrTYb3BJppxMcaZBHQBISHNHY2PpDFlu7qgPHlMN57QjoZsqgXd\nyW2UbvCJJmGvrzHUisRvSbpIYwqKWYpS+5hFxyLTlEqT5x16l/LubAoVKL9iaWsSSUmjJ9EZRgyL\nUNNMR7ysA8rXsFpQX8w5HY9xRWCvSDjwNb6MtDPNolpTDB2msmzKgN9GWkmpzCFOzcntHlNjGeQO\n5QKlX7NpHEM7wImg2hJROVaGiGlJr+V0TgEt2i/Z7F1i6eGiall3a5wHt6o5sil/8ef/k+9rfv1u\nr10/khqs75Tr/ntljx+/zp+tX+Yd4vvqcnrg9C6I2qPmgDmH6pRCn2DUGahTtmbLImm4cJELPeRc\nJpx3M85lyjyO6NwIo0coMyYkY6p0gv8XsHSpKD1l5F1hywAAEC5JREFUrBiU9Dd2S6e4yBQMC6Cg\nZxv+zqZEsNGTSEdCQzqoyUclqapJaUhY7BQKNQSDigYdNTr2pBomGmw0mKixu37vFh0UVgJpbEhC\nIImBJPrdTb1/Wk/j/O7mvgUiSFRIVOjYv56KA2DSg1qxBCwe11ODsqsNol8pjLtIUNxtD/TZCU8T\nlSSiJKBESCWSNZGjZle3tHsM+qJbpUIP1lRAq4BWvm+1x6gOo3qleKs7jA5Y3aGNx6pesV7vSFBU\nfzbwLtgThUfTKUUTDa1YmqCpOkONosJSKUuNpcL1YxIa1U++mgSzqxeyOIxYrBj6SiKDE0OGIhPI\ngQxFqhQpkPD/t3fuMZJU1x3+TlX1Y3bZN2ATWO+CBcTkZYhlEcm24kB4KWZJbFmQ2MYJSmQ7REFO\nFBGhRMiR/3CsWJEVFIJlZBvZZkMcx5PYkU0MjiXLgHmseRpY8CZA1gvLws7CTs9U1T35496eqR6m\nZ5rd6qru3vNJ1XXr9u2u3z1163bdqnNPQ0MklBWagk8rNMTPQkqISIhJaBOvMjgCljwd6pC6Dofd\nHB1N6UjOYcmZjRyzcWFJcjqJ0kkc84lCUyCJ/CTeWJBEiGKQyCGSIjrnXTbIwSlCgmrTu0FqE6cp\njhSni8/NnPoALFl4zf00dz8XT4QMIRPfPqIM4kxIUkhSoZlGtNKIqTSiOR/RTiPanYiprEUra9HM\n2jTyKRquTeLaRLoWkfUQbcQlm8iTjWhjPYn42wvabZ2SE+WvEukhIn0FeAF4BpV5NE7J4xyXKGkj\nYq6ZMN/sfaIcOzgujXljGrP55RYbs7VsyFusd23iqAGthPwER3pCTiZKFvvoSw06bKDDBmYRZol5\nlTXJDOuTl1kTv0LEE8CjiHeIRaYyOCFDyUBywOEkQYlQF5PNJWSzCdmMkKcn4mQb0tzKhqmtbFx7\nFtui8NOQAmkHSXKib+1nywdPpPlzxw3UNx2rqOpNwE3gbw7WoaHdiNm6eU0hZw2weWHrLZUrMgzD\nOLaofIAVfN1voODrLiLTRV/3YXHiz5/ML+w5AFGORsLt+17iZT1MlD7PkyJ849cuIpPjcNEGVE5D\no4gsScL8kf7+muKUOHMkqRKnSvJKzroXUxqdOeJOTjSXo3OObC7HpTmSLwZ58K5e/kC0NWVdnNFu\nzNNopsRxThJnTM3N0pjrIJrjVMkimGskHJqaYqa9hldabWabLWZbLeaaTQ7Gm4KLnneLIAYSRRve\npc0REcIn9F+WeaoXZylTndmwpf7P+LTrnedd4hJNabqUhma08nnW5B0ilyKa+nlUqsQaBpgosQtK\nNMw4cv56JA5/DyY4H8ET/1nAP/lTP48IF4I/OLzbY9iHuG7ar0WVJFci9TaPgdglxBoTazu4t/n9\ndsMdSPjfh8RFxEQkeNeEmIhEhZiINjFrw7wlH41pMXiCn8sUhzlL0UJ0im6Qg0wcGc6740l3INEh\npxs50rc5pzmKQ9Whqjgcs+qYRX1ABXUoGiI7aijr15nmYYiSkalf8jCfyM8r8gPH7uyuhbUEN0iC\n26PDRx9JlQT/F4PrNLhMLvhsLq5VJZwyfp4XIogmIE3Qdd6FUYTga+nX3W3x7bO7rRLKSLQQnENl\nwfczuGZ03UG7+booK1Ifr6IVXHyAOaAji/OkAP/0VFOQvag8g4v8n7K6CFy0+ARxeWJyF+OcA+eQ\nPEPmO+jhVyD1d28lnUeznFTgUNQmi9scjFu04jatpEUjbtGMmiSRd19cq37uWcxxxLKJmEYYnLcW\nJujOA/NuBVn9aIRlvd9ceRjeRjPIX5pj/tlDkzrAeg7YWtg+JeQtV+bZ4CK4AR/swjAMwzB6qOMJ\n1tuB3ar6NICI3ArsAIY+wHr3jt/s2T5xpoMAd3z2k7jHHuGkN/8Sscu9+57LSVzOo6f/ClnSQJ56\nFU0dkjpIFemm593CoADCE5eQbqjSdI62y5ly87TdHO18llZ+iHY2w5R7lXbeoe06tN0ca/OUD267\n2l/MLheGr3vtCOHOMjDTv76KctDlHMhn2Z/Ns18zXnQpB+IOneZhksRxjf4yiltyeVzcpfSsl15I\nv5YE/9ylPPbLDP/evG9hWxcjCvDa28OFd/tIXDhGYx7B/fXRDQ8+yiiQg/YGVhDtk2aZtC7ZlmK5\nxUFZhvfJ13CzAV0cYkY6hWhMlMdEGtHQGAnL5mgvb13zH7yg6/mr9MPMk9DRhHmSEJSksKuuyQue\ndArMXXjyQNb42DdfZsuh5UZPmb9nIn6J8eG7F6YWhvWr0d000sdZ30lY14lpzytJ5mim0Mi6NyyU\nF9YpD5wuNOYdM2uUn75RcKK42M+7O2/7+Vx25mUQCcmWcs/tEeJHwOkicip+IHU58LtLykwDVwI/\nBN4H3LHS/CvDMAzj2KXyOViD+LqH/AU/duBM4PGj3PXxwP5VSx07mD16MXv0YvboxezRSxn22Kaq\nJ5QhpgxE5BLg7/Fj1ptV9ZMi8gngXlWdFpE2cAtwNnAAuLx7o3CF73wBONpJWOPU9sZF67johPHR\nOi46YXy0jotOGB+tlf12jewAawj7vXdUJlSPAmaPXswevZg9ejF79GL2qI5xsvW4aB0XnTA+WsdF\nJ4yP1nHRCeOjtUqddThJDeLrbhiGYRiGYRiGMXbUMcBa8HUXkSbe1326Bh2GYRiGYRiGYRilUvms\nd1XNRORq4Nss+ro/UsGub6pgH+OE2aMXs0cvZo9ezB69mD2qY5xsPS5ax0UnjI/WcdEJ46N1XHTC\n+GitTOdY/NGwYRiGYRiGYRjGOHBMBao2DMMwDMMwDMMYJjbAMgzDMAzDMAzDKImJG2CJyEUi8riI\n7BaRa5d5vyUiO8P7d4vI9upVVsMAtvi4iDwqIg+KyHdFZFsdOqtkNZsUyr1XRFRERj7s6JEyiC1E\n5P2hjTwiIl+pWmOVDHC+vElE7hSRB8I5c0kdOqtCRG4WkedF5OE+74uIfDbY60EROadqjZPOoP3V\nEPe/NbT5bh/wpyH/ehF5TkR2heWSwmf+Muh9XEQurKouIrJHRB4Keu4NeZtF5HYReTKsN4X8vm1X\nRK4M5Z8UkSuHoPPMgt12iciMiFwzKjZd7rwv044i8qvhOO0OnxWOgD46Py0iPwlavi4iG0P+dhGZ\nLdj2xtX09KtziVpLO97ig8bdHfJ3ig8gV5bOnQWNe0RkV8ivzabSv18arXaqqhOz4INmPAWcBjSB\nHwNnLSnzMeDGkL4c2Fm37hpt8W5gTUh/dFJt8XpsEsqtA74P3AW8rW7dNbaP04EHgE1h+8S6ddds\nj5uAj4b0WcCeunUP2SbvAs4BHu7z/iXAfwICnAvcXbfmSVoG7a+GrOEk4JyQXgc8Edr+9cCfL1P+\nrKCzBZwa9MdV1AXYAxy/JO9vgWtD+lrgUyG9bNsFNgNPh/WmkN405GP8M2DbqNh0ufO+TDsC94Sy\nEj57cYk6LwCSkP5UQef2FfqxZfX0q3OJWks73sA/4//4HOBGwu9UGTqXvP93wF/XbVP690sj1U4n\n7QnW24Hdqvq0qs4DtwI7lpTZAXwxpP8FOO9I76CMOKvaQlXvVNXDYfMu/H+STTKDtA+Av8F3zp0q\nxVXMILb4Q+AGVX0JQFWfr1hjlQxiDwXWh/QG4P8q1Fc5qvp94MAKRXYAX1LPXcBGETmpGnXHBIP2\nV0NDVfeq6v0hfQh4DDh5hY/sAG5V1TlV/SmwG1+PuupS/L3/InBZIX+5tnshcLuqHgj93u3ARUPU\ndx7wlKr+zwplKrVpn/O+FDuG99ar6l3qr2K/VPiuo9apqt9R1SxsrnpNs4qefnUuResKvK7jHa5f\nfwN/PXtUWlfSGfbzfuCrK31HFTZdoV8aqXY6aQOsk4FnCtvP8tofg4Uy4UQ8CGypRF21DGKLIlfh\nR+mTzKo2CY+Ot6rqN6sUVgODtI8zgDNE5AcicpeIDPNCo24Gscf1wAdE5FngW8CfVCNtZHm9fYzx\n+hgp+4p3pz8buDtkXR3cbW4uuPr001xFXRT4jojcJyJ/FPLeoKp7Q/pnwBtGQGeRy+m9YB01m3Yp\ny44nh/TS/GHwB/Re05wq3r37v0XknSFvJT396lwmZRzvLcDLhYHlsGz6TmCfqj5ZyKvdpkv6pZFq\np5M2wDKOABH5APA24NN1a6kTEYmAzwB/VreWESHBuwn+OnAF8DkJPu3HKFcAX1DVU/AuB7eENmMY\nE42IHAd8DbhGVWeAfwTeDLwV2It3Haqbd6jqOcDFwB+LyLuKb4Y70SPzvzRhnsylwG0haxRt+hpG\nzY7LISLXARnw5ZC1F3iTqp4NfBz4iois7/f5pQypzmNxvAtcQe/NgNptuky/VOr3Hy2TdnHwHLC1\nsH1KyFu2jIgkeFefFytRVy2D2AIROR+4DrhUVecq0lYXq9lkHfCLwPdEZA/e/3ZaJjPQxSDt41lg\nWlXT4KrwBH7ANYkMYo+r8L7uqOoPgTZwfCXqRpOB+hjjiBkJ+4pIA38R82VV/VcAVd2nqrmqOuBz\nePcl6K956HVR1efC+nng60HTvq7balh33Zxr01ngYuB+Vd0XdI+cTQuUZcfn6HXbK12ziHwY+C3g\n98JFNsHd7sWQvg8/l+mMVfT0q3MplHi8X8S7vCXL1KEUwnf/DrCzoL9Wmy7XL63w/bW000kbYP0I\nOF18RJUm/vH79JIy00A3Usj7gDu6J+GEsaotRORs4J/wg6tJnl/TZUWbqOpBVT1eVber6na8D/el\nqnpvPXKHyiDnyr/hn14hIsfjO8+nqxRZIYPY43/xcyYQkbfgB1gvVKpytJgGPhQiNJ0LHCy4ZxhH\nzyBtcqiEeRefBx5T1c8U8otz7X4b6EYdmwYuFx+t91T8DZl7GHJdRGStiKzrpvHBDh6m9/f+SuAb\nBZ3Ltd1vAxeIyKbgsnVByBsGPU8ERs2mSyjFjuG9GRE5N7StDxW+66gJbux/gf/dPlzIP0FE4pA+\nDW/Dp1fR06/OZWkt5XiH69c78dezQ9EKnA/8RFUX3ObqtGm/fmmF76+nneoRRkUZ1QXvuvMEfjR9\nXcj7BP6EA39RdBt+4uA9wGl1a67RFv8F7AN2hWW6bs1122RJ2e8xoVEEB2wfgneZfBR4iBClaFKX\nAexxFvADfPSmXcAFdWsesj2+incDSfFPM68CPgJ8pNA+bgj2emiSz5Uaj8Fr2mTF+38H3s3mwcLv\nxCXALeGYP4i/eDmp8Jnrgt7HKUTeGmZd8JHVfhyWRwrn7xbgu8CT4fduc8jv23bxc3d2h+X3h2TX\ntfgnDxsKeSNh0z7nfWl2xE9HeDh85h8AKVHnbvycmm5b7UaMfm9oF7uA+4H3rKanX51L1Fra8Q7t\n/55Q/9uAVlk6Q/4XCP1+oWxtNqV/vzRS7bRbacMwDMMwDMMwDOMomTQXQcMwDMMwDMMwjNqwAZZh\nGIZhGIZhGEZJ2ADLMAzDMAzDMAyjJGyAZRiGYRiGYRiGURI2wDIMwzAMwzAMwygJG2AZhmEYhmEY\nhmGUhA2wDMMwDMMwDMMwSuL/AZEHApTLsacnAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fad5fd73f98>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "pm.plots.traceplot(trace)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### III. b) YN implementation - Labeling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Using the parameters estimation from the Training Stage\n",
    "# as prior parameter for the Labeling Stage\n",
    "\n",
    "# This is one of the most important parts of the proposed model\n",
    "theta_val = trace[theta][-5000::20].mean(axis=0)\n",
    "theta_sd = trace[theta][-5000::20].std(axis=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 3000/3000 [01:10<00:00, 42.68it/s]\n"
     ]
    }
   ],
   "source": [
    "with pm.Model() as model2: \n",
    "    \n",
    "    pi = pm.Dirichlet('pi', p)\n",
    "    z_unknown = pm.Categorical('z_unknown', p=pi, shape=N_test)\n",
    "    new_theta = pm.Beta('new_theta', mu=theta_val, sd=theta_sd, shape=(J, K, K))        \n",
    "\n",
    "    # EQUATION (3.2)\n",
    "    def likelihood_Pzt_r(value, zt=z_unknown, theta=new_theta, pzs=pi):\n",
    "        l = np.log(theta[:, zt,:]**value[:,:,:,0] * (1-theta[:,zt,:])**(value[:,:,:,1]))\n",
    "        return l.sum()\n",
    "    \n",
    "    r = pm.DensityDist('r', likelihood_Pzt_r, observed=(r_test))\n",
    "\n",
    "with model2:   \n",
    "    step1 = pm.Metropolis([new_theta, pi])\n",
    "    step2 = pm.CategoricalGibbsMetropolis([z_unknown])\n",
    "    \n",
    "    # for full convergence njobs should be about 10.\n",
    "    # for sppiding up the demo we use njobs=1\n",
    "    trace2 = pm.sample(3000, step=[step1, step2], njobs=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Credibility part TRUE proportion distribution [ 0.34146341  0.2195122   0.17073171  0.26829268]\n",
      "Labeling part TRUE proportion distribution [ 0.2375  0.2125  0.225   0.325 ]\n",
      "Labeling part LEARNED proportion distribution [ 0.14760714  0.246984    0.28418146  0.32122739]\n"
     ]
    }
   ],
   "source": [
    "print(\"Credibility part TRUE proportion distribution {}\".format(p))\n",
    "print(\"Labeling part TRUE proportion distribution {}\".format(p_test))\n",
    "print(\"Labeling part LEARNED proportion distribution {}\".format(trace2[pi][-100::10].mean(axis=0)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "# MAP\n",
    "v, f = stats.mode(trace2[z_unknown][-500::10])\n",
    "result = v[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 0.684  0.158  0.158  0.   ]\n",
      " [ 0.     1.     0.     0.   ]\n",
      " [ 0.     0.     1.     0.   ]\n",
      " [ 0.     0.     0.     1.   ]]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAASAAAAElCAYAAABETj8zAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fad38ae7dd8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Acurracy score 0.925\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<table><tr><td>Metrics / Classes</td><td>CEP</td><td>EB</td><td>LPV</td><td>RRLYR</td></tr><tr><td>Precision score</td><td>1.0</td><td>0.85</td><td>0.857</td><td>1.0</td></tr><tr><td>Recall score</td><td>0.684</td><td>1.0</td><td>1.0</td><td>1.0</td></tr><tr><td>F score</td><td>0.813</td><td>0.919</td><td>0.923</td><td>1.0</td></tr></table>"
      ],
      "text/plain": [
       "[['Metrics / Classes', 'CEP', 'EB', 'LPV', 'RRLYR'],\n",
       " ['Precision score', 1.0, 0.84999999999999998, 0.85699999999999998, 1.0],\n",
       " ['Recall score', 0.68400000000000005, 1.0, 1.0, 1.0],\n",
       " ['F score',\n",
       "  0.81299999999999994,\n",
       "  0.91900000000000004,\n",
       "  0.92300000000000004,\n",
       "  1.0]]"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "draw_metrics(zt_test, result, [classes_dict[i] for i in range(K)])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# IV. ABCD - PYMC IMPLEMENTATION"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### IV. a) ABCD implementation - Credibility"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "votes_abc = pd.read_csv(selected_folder+'votes_abcd.csv')\n",
    "votes_abc = votes_abc[votes_abc.expert_id.isin(experts)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Generatting the ABCD matrix of votes from the source data\n",
    "def matrix_votes_abcd(J, N, K, o_columns, votes):\n",
    "    r_ = np.zeros((J, N, K))\n",
    "    for i in range(o_columns.shape[0]):\n",
    "        instance = o_columns.iloc[i]\n",
    "        _votes = votes[votes.object_id == instance.catalina_id]\n",
    "        for v in range(_votes.shape[0]):\n",
    "            vote = _votes.iloc[v]\n",
    "            r_[inv_experts_id[vote.expert_id], i,  inv_classes_dict[vote.value]] = 1\n",
    "    return r_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Getting the votes for the users KNOWN LABELS - TRAINING PART\n",
    "r_true = matrix_votes_abcd(J, N, K, o_columns[o_columns._training], votes_abc)  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "crowd_J = J"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_classes(r):\n",
    "    return np.array([[_.tolist().index(1) for _ in expert] for expert in r])\n",
    "\n",
    "crowd_r = get_classes(r_true)\n",
    "\n",
    "data = np.zeros((crowd_J, K, K))\n",
    "for i in range(crowd_J):\n",
    "    data[i] = confusion_matrix(zt, crowd_r[i])\n",
    "    \n",
    "norm = np.array((data.reshape((crowd_J*K,K))/np.matrix(data.sum(2).reshape(crowd_J*K)).T)).reshape((crowd_J,K,K))\n",
    "obs = data.reshape((crowd_J*K,K)).sum(1).reshape((crowd_J,K))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 20000/20000 [00:05<00:00, 3523.49it/s]\n"
     ]
    }
   ],
   "source": [
    "with pm.Model() as model:\n",
    "    theta = pm.Dirichlet('theta', np.ones(K)/K, shape=(crowd_J*K,K))\n",
    "    likel = pm.Multinomial('likel', obs.reshape((crowd_J*K)), theta, \n",
    "                           observed=data.reshape((crowd_J*K,K)), shape=(crowd_J*K,K))\n",
    "\n",
    "with model:\n",
    "    step = pm.Metropolis([theta,likel])\n",
    "    trace = pm.sample(20000, step)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### IV. b) ABCD implementation - Labeling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "r_test = matrix_votes_abcd(J, N_test, K, o_columns[~o_columns._training], votes_abc)  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "theta_val = trace[theta][-5000::20].mean(axis=0).reshape((crowd_J,K,K))\n",
    "theta_val= np.array((theta_val.reshape((crowd_J*K,K))/np.matrix(theta_val.sum(2).reshape(crowd_J*K)).T)).reshape((crowd_J,K,K))\n",
    "theta_sd = trace[theta][-5000::20].std(axis=0).reshape((crowd_J,K,K))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "crowd_r_u = get_classes(r_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "crowd_r_u = get_classes(r_test)\n",
    "\n",
    "data_test = np.zeros((crowd_J, K, K))\n",
    "for i in range(crowd_J):\n",
    "    data_test[i] = confusion_matrix(zt_test, crowd_r_u[i])\n",
    "data_test\n",
    "\n",
    "norm_test = np.array((data_test.reshape((crowd_J*K,K))/np.matrix(data_test.sum(2).reshape(crowd_J*K)).T)).reshape((crowd_J,K,K))\n",
    "obs_test = data_test.reshape((crowd_J*K,K)).sum(1).reshape((crowd_J,K))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 5000/5000 [02:06<00:00, 39.47it/s]\n"
     ]
    }
   ],
   "source": [
    "with pm.Model() as model2: \n",
    "    pi = pm.Dirichlet('pi', p)\n",
    "    \n",
    "    z_unknown = pm.Categorical('z_unknown', p=pi, shape=(N_test))\n",
    "\n",
    "    new_theta = pm.Dirichlet('theta', theta_val, shape=(crowd_J,K,K))\n",
    "\n",
    "    def likelihood_Pzt_r(value, zt=z_unknown, theta=new_theta, pzs=pi):\n",
    "        l = np.log(theta[:, zt,:]**value)\n",
    "        return l.sum()\n",
    "\n",
    "    r = pm.DensityDist('r', likelihood_Pzt_r, observed=(r_test))\n",
    "\n",
    "with model2:\n",
    "    step0 = pm.Slice(pi)\n",
    "    step1 = pm.Metropolis([new_theta])\n",
    "    step2 = pm.CategoricalGibbsMetropolis([z_unknown])\n",
    "    trace2 = pm.sample(5000, step=[step0, step1, step2], njobs=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([ 0.34146341,  0.2195122 ,  0.17073171,  0.26829268]),\n",
       " array([ 0.2375,  0.2125,  0.225 ,  0.325 ]),\n",
       " array([ 0.14042533,  0.26713698,  0.26780912,  0.32462857]))"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "p_test = Ns_test/ Ns_test.sum()\n",
    "p, p_test, trace2[pi][1000::10].mean(axis=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "_trace = trace2.get_values('z_unknown', chains=0)\n",
    "v, f = stats.mode(_trace[1000::10], 0)\n",
    "result = v.T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 0.579  0.211  0.158  0.053]\n",
      " [ 0.     1.     0.     0.   ]\n",
      " [ 0.     0.     1.     0.   ]\n",
      " [ 0.     0.     0.     1.   ]]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAASAAAAElCAYAAABETj8zAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fad4ff4d2b0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Acurracy score 0.9\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<table><tr><td>Metrics / Classes</td><td>CEP</td><td>EB</td><td>LPV</td><td>RRLYR</td></tr><tr><td>Precision score</td><td>1.0</td><td>0.81</td><td>0.857</td><td>0.963</td></tr><tr><td>Recall score</td><td>0.579</td><td>1.0</td><td>1.0</td><td>1.0</td></tr><tr><td>F score</td><td>0.733</td><td>0.895</td><td>0.923</td><td>0.981</td></tr></table>"
      ],
      "text/plain": [
       "[['Metrics / Classes', 'CEP', 'EB', 'LPV', 'RRLYR'],\n",
       " ['Precision score',\n",
       "  1.0,\n",
       "  0.81000000000000005,\n",
       "  0.85699999999999998,\n",
       "  0.96299999999999997],\n",
       " ['Recall score', 0.57899999999999996, 1.0, 1.0, 1.0],\n",
       " ['F score',\n",
       "  0.73299999999999998,\n",
       "  0.89500000000000002,\n",
       "  0.92300000000000004,\n",
       "  0.98099999999999998]]"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "draw_metrics(zt_test, result, [classes_dict[i] for i in range(K)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}