Update Jupyter notebook analysis

[hmakelin]

New accuracy analysis using pyulog (px4tools seems to no longer be maintained). Some new insight into how flight altitude affects accuracy of estimates.

As can be seen from the histograms - estimates are all biased, some potential causes:

• Poor test design: drone and camera facing only east with slight perspective in camera causes bias in easting/northing directions
• Comparing AGL altitude against AMSL altitude causes bias in altitude/z-axis direction as simulation starting location is a few meters above sea level
• Potential undiscovered issues with implementation, esp. coordinate frame conversions