feat(audit): VW-Cookie-Bug-Fix + P101/P102 Cookie-Library-Mismatch-Findings

VW-Bug B1: extract_vendors_via_llm hatte max_text_chars=12000 -> bei
VW-Cookie-Doc (60k chars, 100 Cookies in Tabelle) wurden 80% abgeschnitten,
LLM extrahierte nur 1 Vendor. Fix: max_text_chars=50000, num_predict
6000->16000 fuer mehr Vendor-Output, Ollama-Timeout 120s->420s.

P101 Aggregator-Script (backend-compliance/scripts/cookie_library_enrich.py)
geht alle compliance_check_snapshots durch und extrahiert (cookie_name,
declared_category, observed_sites). Erste Auswertung ueber 8 Snapshots:
101 unique Cookies, 47 in Library, 54 unbekannt, 18 Mismatches.

P102 Cookie-Klassifikations-Pruefung als Mail-Block. Vergleicht
Site-deklarierte Kategorie vs Library + Vendor-Doku. HIGH wenn Library
sagt 'marketing' aber Site als 'essential'/'statistics' deklariert
(faktische Drittland-/Werbe-Verarbeitung versteckt). MEDIUM sonst.
In agent_compliance_check_routes Mail-Komposition + Replay-Pipeline
eingebaut.

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

backend-compliance/compliance/services/check_replay.py

@@ -116,6 +116,29 @@ def replay_from_snapshot(
     except Exception as e:
         logger.warning("Replay: vvt failed: %s", e)
 
+    # P102: Cookie-Klassifikations-Pruefung
+    try:
+        from compliance.services.cookie_library_mismatch import (
+            detect_mismatches, build_mismatch_block_html,
+        )
+        cookies_seen: list[str] = []
+        for ph in (banner_result.get("phases") or {}).values():
+            if isinstance(ph, dict):
+                for ck in (ph.get("cookies") or []):
+                    if isinstance(ck, str):
+                        cookies_seen.append(ck)
+                    elif isinstance(ck, dict) and ck.get("name"):
+                        cookies_seen.append(ck["name"])
+        doc_for_check = doc_texts.get("cookie") or doc_texts.get("dse") or ""
+        if cookies_seen and doc_for_check:
+            mm = detect_mismatches(db, cookies_seen, doc_for_check)
+            if mm:
+                mm_html = build_mismatch_block_html(mm)
+                parts.append(mm_html)
+                section_sizes["library_mismatch"] = len(mm_html)
+    except Exception as e:
+        logger.warning("Replay: mismatch block failed: %s", e)
+
     full_html = "".join(parts)
 
     result = {