feat(cra): Datenblatt→Grenzen-Extraktor (hybrid, lokales 35B)

Hybrid-Extraktion Datenblatt → IACE Grenzen (ISO 12100): deterministischer
Detektor (Schnittstellen/Einheiten per Regex) + lokales 35B via llm_cascade
(Qwen-lokal-first) fuer die semantische Zuordnung auf die echten LimitsFormData-
Keys. Nichts erfinden: Feld nicht im Text → leer + Quellen-Zitat je Feld.
Essenzielle ISO-12100-Felder, die leer bleiben → gezielte Rückfragen
(foreseeable_misuses, person_groups, qualification, temporal_limits …).
Endpoint POST /api/v1/cra/extract-datasheet. 13 Tests gruen (reine Teile).

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

backend-compliance/compliance/api/cra_assess_routes.py

@@ -18,6 +18,7 @@ from compliance.services.cra_finding_mapper import assess_findings_payload
 from compliance.services.cra_applicability import (
     compute_verdict, compute_machinery_verdict, maturity as evidence_maturity, MACHINE_INTEGRATOR,
 )
+from compliance.services.cra_datasheet_extractor import extract_grenzen
 from compliance.services.scanner_mcp_client import fetch_findings
 from compliance.services.cra_snapshot_store import save_snapshot, list_snapshots, get_snapshot
 from compliance.services.cra_use_case_controls import enrich_findings_with_breadth
@@ -137,6 +138,18 @@ async def assess_from_scanner(body: ScannerPullRequest):
     return result
 
 
+class DatasheetRequest(BaseModel):
+    text: str = ""
+
+
+@router.post("/extract-datasheet")
+async def extract_datasheet(body: DatasheetRequest):
+    """Datasheet text -> IACE 'Grenzen' draft (limits + provenance) + the
+    essential ISO-12100 fields still missing as targeted follow-up questions.
+    Hybrid: deterministic interface/unit detector + local 35B (llm_cascade)."""
+    return await extract_grenzen(body.text)
+
+
 @router.get("/scanner-repos")
 async def scanner_repos():
     """Distinct repo_ids the scanner has findings for, so the UI can pick which