feat: add metadata extraction and synthetic keyword detection for images and videos

refactor: clean up model roster comments and remove broken model references

fix: update Dockerfile dependencies, remove audio processing, and replace models

feat: enhance model capabilities and update dependencies

Refresh Gemini candidate order to current Gemini 3 model codes

Update model stack, add audio stream extraction, and harden inference wiring

fix(space): set PYTHONPATH and prefer Gemini 3.1 with 2.5 fallback

Initial commit