fix: use structured report_json and show on-page autofill status

fix: Robust JSON parsing and field mapping for report generation

Fix AI report: enforce exact 10-key JSON schema in backend prompt, fix duplicate aceticFindings key

Merge remote HF Space changes, keeping local AI report fixes

Fix AI report: JSON field mapping, local dev proxy, dotenv loading, fuller exam data sent to LLM

Fix restart loop: implement lazy loading for YOLO models

Fix AI report generation JSON mapping and report UI fields

Fix report structure: require all sections (biopsy, findings, diagnostics, management) with explicit formatting

Prioritize Gemini 2.5 models and cache quota-blocked model fallbacks

Add /api/health endpoint alias for HF Spaces

Use Gemini list_models dynamically for chat/report model selection

Add model listing and try more compatible model names for v1beta API

Fix Gemini API key authentication for HF Spaces

Add model fallback and better error handling for Gemini API

Add comprehensive LLM endpoints and enhance FastAPI backend

Fix import error: use relative import for inference module

Update application with new features and components

Remove ultralytics version pin to use latest compatible

Patch Segment head and downgrade to ultralytics 7.3.23

Downgrade ultralytics to 8.0.227 for model compatibility

Fix ultralytics model loading with CPU device and error handling

Add detailed error logging to infer-aw-contour endpoint

Fix backend import for Space

Run FastAPI in Space and use relative API URLs

Update backend and UI; add models