Fix GRPO all-zero rewards: parse JSON actions from completions

Fix GRPOConfig: use fp16 for T4 GPU, remove deprecated logging_dir

Fix GRPOConfig: remove chat_template_kwargs unsupported by Unsloth

Add demo script, API endpoints, and training notebook baselines

Fix training notebook to use WebSocket client for stateful sessions

Deploy to HF Spaces + add TRL/Unsloth training notebook