Fix: Set TF_USE_LEGACY_KERAS=1 for Keras 3 compatibility

Fix: Update Dockerfile for HF Spaces (Port 7860, Permissions)

Fix: Explicit HEAD support and switch to port 10000 for Render

Critical: Aggressive memory fix for Render 512MB RAM + Port detection fix

Optimize: Pre-download AI models in Docker for faster startup

Fix libgl1 dependency for Render

Setup for deployment