also implies that agents are drawn in their old locations, ie from previous timestep

moved test_play.py to use player.py rather than contrived tkplay.py
trying to improve logic around gl.show() show param to renderEnv

moved test_play.py to use player.py rather than contrived tkplay.py
trying to improve logic around gl.show() show param to renderEnv

moved test_player.py to using PIL and tkinter instead of QT

Added load and save to env (left GridTransitionMap load and save for now)

Added PIL graphics - 5-10 times faster than MPL
move adaptColor into GraphicsLayer
change scatter size arg to s like MPL
move clf, get_cmap up into GraphicsLayer
added boolean "train" arg to ddqn step() to make running faster in player / editor
add plotTreeObs - not finished.
improved pixel scaling in editor (I hope)
Made the agent background steps update a progress bar.

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.