chore(qa): add PDF-based control QA scripts and results

QA pipeline that matches control source_original_text directly against
original PDF documents to verify article/paragraph assignments. Covers
backfill, dedup, source normalization, Qdrant cleanup, and prod sync.

Key results (2026-03-20):
- 4,110/7,943 controls matched to PDF (100% for major EU regs)
- 3,366 article corrections, 705 new assignments
- 1,290 controls from Erwägungsgründe (preamble) identified
- 779 controls from Anhänge (annexes) identified

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

scripts/qa/delete_gpsr_prod.py

@@ -0,0 +1,27 @@
+"""Delete eu_2023_988 duplicate from production Qdrant."""
+import httpx
+
+PROD_URL = "https://qdrant-dev.breakpilot.ai"
+HEADERS = {"api-key": "z9cKbT74vl1aKPD1QGIlKWfET47VH93u"}
+
+# Delete
+resp = httpx.post(
+    f"{PROD_URL}/collections/bp_compliance_ce/points/delete",
+    json={"filter": {"must": [{"key": "regulation_id", "match": {"value": "eu_2023_988"}}]}},
+    headers=HEADERS, timeout=60,
+)
+print(f"Delete status: {resp.json().get('status')}")
+
+# Verify
+resp2 = httpx.post(
+    f"{PROD_URL}/collections/bp_compliance_ce/points/count",
+    json={"filter": {"must": [{"key": "regulation_id", "match": {"value": "eu_2023_988"}}]}, "exact": True},
+    headers=HEADERS, timeout=15,
+)
+remaining = resp2.json().get("result", {}).get("count", 0)
+print(f"Remaining: {remaining}")
+
+# Total
+resp3 = httpx.get(f"{PROD_URL}/collections/bp_compliance_ce", headers=HEADERS, timeout=10)
+total = resp3.json().get("result", {}).get("points_count", "?")
+print(f"Total points: {total}")