Added some track waypoint analysis to help reward functions - 2nd attempt

[fmacrae]

My racing name is drunken-monkey
This change has been submitted as part of the AWS DeepRacer Community Log Analysis Challenge. To join the community go to http://joindeepracing.io

What I built:
Building a starter for track analysis to create outputs we can put into reward function logic.

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[fmacrae]

Added type hints and docstrings for some of the track utils too. Hoping the Analysis can be used to overlay log results so we can colour code where the robot is doing something counter intuitive, so should be OK for the log analysis challenge.... :)

[fmacrae]

Screenshot of the output visualization of the Canada track with waypoint analysis done and colour coding with legend. Tested analysis on all current tracks and works fairly well. Data points produced are available as a numpy array and can either feed into log analysis to compare track position to average action activations. Basically you don't want too much high throttle on a hairpin or excessive steering on a straight so hoping we can flag ill advised actions taken by the agents during training and evaluation for quick debug. Also added type hints and doc strings to most of the track_utils.py file to help future users and to show how to document the python files in case anyone else new to python wants to start helping there too.