refactor: multi-treatment logic, deep audit fixes for ML shapes and test assertions, security updates

Phase 2 Deep Dive Fixes: Schema updates, race conditions, N+1 queries, and ML pipeline bounds checking

Fix unicode bug in models.py and add os.remove exception handler in views.py

Deployment readiness fixes (Security & ML Robustness)

Remove forced company linking and add seed commands to Dockerfile startup

Add auto-admin script and route DB/Media to persistent storage

Increase CONFIDENCE_THRESHOLD to 0.75

Remove confidence inflation logic

Fix unterminated string literal in LLM prompt and remove confidence inflation logic

Fix confidence inflation bug in predict_disease that bypassed LLM for single-class species

Inject new user-provided Gemini/Groq prompt with advanced reasoning steps

Force Django to serve media files in production (DEBUG=False)

Preload YOLO model during docker build to prevent Gunicorn timeout

Add debugging output and timeout to Docker CMD

Install OpenCV system dependencies (libgl, libxcb) for YOLO

Add ultralytics and cpu torch to fix missing YOLO dependencies

Add huggingface_hub required for Keras 3 model loading

Fix CSRF_TRUSTED_ORIGINS for Hugging Face domains

Add missing geopy and beautifulsoup4

Optimize requirements for blazing fast CPU deployment

Add required Hugging Face README config

Initial Deploy to HF