v1.0 - Algortihm Generation and Deep RL training

This version implements automatic algorithm generation and symmetry detection for puzzles. It also includes tools to train, test and analyze deep reinforcement learning agents for solving general twisty puzzles.
This version includes many new puzzles, some ready-to-use trained neural networks for solving six different puzzles and several new CLI commands to improve the user experience.

v0.2.3 V-tables, better animations

This version controls 3D browser based animations via a command-line interface. This interface does not have any auto-completion.
Supported AIs are only table based: Q-tables and (more efficient) V-tables.
While code for Neural network based AIs exists, they don't seem to learn anything -> code has been disabled.

After this version, older version will no longer be included in the repository. They were a relic of me not trusting git.