Investigate replacing pygrib with cfgrib(probably with xarray) for the ingest processes

[randytpierce]

Describe what needs to be done here.
The Pygrib python package has been having build and install problems so we need to find out if cfgrib is better supported and works as well. This is a blocker for the ingest work.

[randytpierce]

Reference repo https://github.com/ecmwf/cfgrib. This is very possible and now work is proceeding on making this happen per our discussion in the vxIngest meeting July 25th.
Here are a few notes for processing a test model output file with xarray engine='cfgrib'...
python
import xarray
import cfgrib

# how to get surface variables (level 2 meter typeOfLevel heightAboveGround)...
# ds2 = xr.open_dataset('/opt/data/grib2_to_cb/input_files/2125214000000',  engine='cfgrib',backend_kwargs={'filter_by_keys':{'typeOfLevel': 'heightAboveGround', 'stepType':'instant', 'level':2}})
# >>> sorted(ds2.variables  ['d2m', 'heightAboveGround', 'latitude', 'longitude', 'pt', 'r2', 'sh2', 'step', 't2m', 'time', 'valid_time']
# ds2.variables['r2']
# ds2.variables['r2'].values
# How to get wind data at 10 m
# >>> dsh = xr.open_dataset('/opt/data/grib2_to_cb/input_files/2125214000000',  engine='cfgrib',backend_kwargs={'filter_by_keys':{'typeOfLevel': 'heightAboveGround', 'stepType':'instant', 'level':10}})
# how to get ceiling data
# >>> dsc = xr.open_dataset('/opt/data/grib2_to_cb/input_files/2125214000000',  engine='cfgrib',backend_kwargs={'filter_by_keys':{'typeOfLevel': 'cloudCeiling', 'stepType':'instant'}})
# sorted(dsc.variables)
# dsc.variables['cloudCeiling'].dims
# dsc.variables['cloudCeiling'].data
# dsc.coords['latitude']
# dsc.coords['longitude']

[randytpierce]

We have demonstrated that we can make this work and have decided as a team to implement cfgrib via xarray so I am closing this issue.