feat(maker): AI parsing schede tecniche PDF via OpenRouter

- Nuovo servizio ai_service.py: estrae testo da PDF (pdfplumber) + analisi AI (OpenRouter)
- Endpoint POST /api/recipes/{id}/parse-technical-sheet con validazione file
- UI: bottone "Importa da PDF" nel task editor con modale upload + preview editabile
- Task suggeriti modificabili/rimovibili prima della creazione bulk
- Config: OPENROUTER_API_KEY e OPENROUTER_MODEL in .env
- Dipendenze: pdfplumber + httpx aggiunti a server deps

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

src/backend/api/routers/recipes.py

@@ -1,5 +1,7 @@
-"""Recipe router - CRUD, versioning, barcode lookup."""
-from fastapi import APIRouter, Depends, Query
+"""Recipe router - CRUD, versioning, barcode lookup, AI parsing."""
+import logging
+
+from fastapi import APIRouter, Depends, File, HTTPException, Query, UploadFile, status
 from sqlalchemy.ext.asyncio import AsyncSession
 
 from src.backend.database import get_db
@@ -164,3 +166,31 @@ async def get_measurement_count(
     """
     count = await recipe_service.get_measurement_count(db, recipe_id, version_number)
     return {"recipe_id": recipe_id, "version_number": version_number, "measurement_count": count}
+
+
+@router.post("/{recipe_id}/parse-technical-sheet")
+async def parse_technical_sheet(
+    recipe_id: int,
+    file: UploadFile = File(...),
+    _user: User = Depends(require_maker),
+    db: AsyncSession = Depends(get_db),
+):
+    """Parse a PDF technical sheet using AI and return suggested tasks."""
+    from src.backend.services import ai_service
+
+    if not file.content_type or "pdf" not in file.content_type:
+        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail="Il file deve essere un PDF")
+
+    pdf_bytes = await file.read()
+    if len(pdf_bytes) > 20 * 1024 * 1024:
+        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail="File troppo grande (max 20MB)")
+
+    try:
+        tasks = await ai_service.parse_technical_sheet(pdf_bytes)
+    except ValueError as e:
+        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=str(e))
+    except Exception as e:
+        logging.getLogger(__name__).error("AI parsing error: %s", e)
+        raise HTTPException(status_code=status.HTTP_502_BAD_GATEWAY, detail="Errore nel servizio AI")
+
+    return {"recipe_id": recipe_id, "suggested_tasks": tasks}