TorchLean RL Runtime Facade

RL environment, algorithm, replay, PPO, Gymnasium, and training names.

Reinforcement-learning APIs used by the executable examples.

The public namespace mirrors the shape of the checked RL stack:

• rl.core, rl.mdp, and friends expose the mathematical objects,
• rl.replay, rl.dqn, and rl.policy expose runtime update formulas,
• rl.boundary, rl.gym, and rl.eval expose the executable trust-boundary APIs,
• rl.cli keeps runnable examples on one shared command-line convention.