Improved multi-robot rendering capabilities

[GustavoSilvera]

• Improved capability to render multiple robot algorithms on the same episode
• Added run_multi_robot.sh script to automate running several different robot algorithms on the same episode
• Fixed the environment.yml to install the barebones requirements for creating a conda installation
• Updated documentation to docs/
• Updated some params allowing for more consistent rendering
• Improved performance of legend plotting

[GustavoSilvera]

This pr does not include any joysticks besides what was already included (sampling & random planner). So yes, this will raise a NotImplementedError for other joysticks unless they get installed.

[GustavoSilvera]

Putting this PR on hold for a while. It improves some non-critical functionality for users and we want to run more tests before merging.

[GustavoSilvera]

Bumping this PR. It will probably help with a lot of the installation issues from earlier.

@ajdroid wanna run the tests and make sure it looks good

[ajdroid]

Needed to change the opencv install version to 4.2.* in multi-robot-rendering/environment.yml

Had an issue with the pyopenGL similar to: https://github.com/MPI-IS/mesh/issues/23

File "/usr/lib/python3/dist-packages/OpenGL/raw/GLES2/_types.py", line 5, in _error_checker = _ErrorChecker( _p, _p.GLES2.glGetError )

Was able to bypass by setting _error_checker=None in that file. Can you reproduce @GustavoSilvera ?

After this workaround

[GustavoSilvera]

I can confirm that upgrading the opencv to 4.2.* still maintains fully-functional unit tests. But I did not need to perform the _error_checker=None workaround.

I also recently added some changes to address #17 and #18 by simplifying the installation process