metrics: profit first, then service level, then fill rate

Remove PPO MLP agent, update README

fix: let Gradio auto-select port locally, fix via env vars only

fix: use 127.0.0.1 locally, 0.0.0.0 in Docker via env var

fix: Dockerfile runs Gradio app, not FastAPI server

fix: fall back to HF_TOKEN env var for Inference API auth

Added trained PPO model + app.py UI changes for HF Spaces

Add P&L reward function, daily spoilage, stochastic lead time, and reward visualization