Koopman-Assisted Reinforcement Learning Release

This release encapsulates the codebase and how it was used for the paper called "Koopman-Assisted Reinforcement Learning", most specifically introducing the 2 new Koopman extensions

1. Soft Actor Koopman-Critic
2. Soft Koopman Value Iteration

In addition the code permits for the automatic tuning of hyper parameters using the standardized Ray Tune interface.