Applied workaround for spurious matplotlib console output

https://pep8.readthedocs.io/en/latest/intro.html#error-codes

Qt render

See merge request flatland/flatland!11

introduced a simple graphics layer so same code
can drive both matplotlib and pyqt

loads a pre-saved model from Nets (not sure if working)
added some frame, time, episode text info to RenderEnv (still in matplotlib)
added some flake8 ignores to tox.ini - mostly hanging index-related.

to make it easier to run the example

add saved rail env transitionmap data env-data/tests/test1.npy and update basic-env.png to match it.

Global observation builder

See merge request flatland/flatland!10