Reinforcement learning code that allows one to easily create new agents and environments to test and play.