fix(extraction): סינון cited_only מתור/מוני החילוץ (#140)

31 שורות case_law עם source_kind='cited_only' (ציטוט-בלבד, ללא full_text/chunks)
נושאות halacha_extraction_status='pending' רק כברירת-מחדל ומזהמות את מונה ה-pending
ובמתזמר/בדף-התפעול — אין להן מה לחלץ.

תיקון (G1 — תיקון-במקור, G2 — מסנן יחיד משותף):
- db.EXTRACTION_ELIGIBLE_PREDICATE — מקור-אמת יחיד ל"שורה ברת-חילוץ" (source_kind
  <> 'cited_only' AND יש precedent_chunks). מוחל ב-list_pending_extraction_requests;
  #139 יעשה בו שימוש-חוזר ל-reconcile (אותו כלל, לא כפול).
- מוני-snapshot מסננים cited_only: halacha_drain_supervisor.db_snapshot,
  web/app.py meta+hal_ext (GROUP BY status).
- reconcile_metadata_status.py מורחב לכסות גם את תור-ההלכות: cited_only→'skipped'
  (אותו terminal-state כמו צד-המטא, תור-תאום, G2). בוצע על ה-DB החי: 31 הועברו
  ל-'skipped' (metadata כבר היה מיושב — אידמפוטנטי). התפלגות-אחרי: halacha
  pending=9 (עבודה אמיתית), skipped=31, completed=309.

בדיקות: test_extraction_queue_eligibility (predicate + list_pending מחיל אותו,
שני ה-kinds). כל 345 בדיקות mcp עוברות. guards נקיים.

Invariants: G1 (terminal-state אמיתי במקור), G2 (predicate יחיד, ללא תור מקביל),
INV-DM1 (stub לא-searchable אינו מועמד-חילוץ), G12 (leak-guard נקי).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

scripts/reconcile_metadata_status.py

@@ -1,4 +1,4 @@
-"""Reconcile stale ``metadata_extraction_status='pending'`` rows (G1).
+"""Reconcile stale ``*_extraction_status='pending'`` rows (G1).
 
 The column defaults to 'pending', but only ``source_kind='external_upload'``
 rows with extractable text genuinely need the Gemini metadata drain. Internal
@@ -13,6 +13,10 @@ This settles each row to a truthful terminal state at the source:
   - external_upload w/ text but missing name/summary → stamp requested_at (real work → drain picks it up)
   - cited_only (no text) → 'skipped' (terminal; nothing to extract)
 
+Covers BOTH extraction queues (#140): the cited_only→'skipped' settle is applied
+to ``halacha_extraction_status`` as well as ``metadata_extraction_status`` — same
+phantom-backlog fix on the twin queue, one reconcile script (G2).
+
 Idempotent and re-runnable (a healthy DB reports all-zero). The companion source
 fix lives in db.create_internal_committee_decision (inserts 'completed' directly)
 so internal rows never re-enter this state.
@@ -57,6 +61,15 @@ async def main() -> int:
         "WHERE source_kind = 'cited_only' "
         "AND metadata_extraction_status = 'pending'"
     )
+    # Halacha side (#140): cited_only stubs inherit DEFAULT 'pending' but have no
+    # text/chunks to extract holdings from — settle them to the same terminal
+    # 'skipped' the metadata side uses, so they stop inflating the halacha
+    # pending counter / supervisor snapshot. Same source-fix, one reconcile (G2).
+    cited_hal = await pool.execute(
+        "UPDATE case_law SET halacha_extraction_status = 'skipped' "
+        "WHERE source_kind = 'cited_only' "
+        "AND halacha_extraction_status = 'pending'"
+    )
 
     def n(tag: str) -> str:
         try:
@@ -64,10 +77,11 @@ async def main() -> int:
         except (AttributeError, IndexError):
             return "?"
 
-    print(f"internal_committee  → completed : {n(internal)}")
-    print(f"external_upload     → completed : {n(external_done)}")
-    print(f"external_upload     → requeued  : {n(external_requeued)}")
-    print(f"cited_only          → skipped   : {n(cited)}")
+    print(f"internal_committee  → completed       : {n(internal)}")
+    print(f"external_upload     → completed       : {n(external_done)}")
+    print(f"external_upload     → requeued        : {n(external_requeued)}")
+    print(f"cited_only metadata → skipped         : {n(cited)}")
+    print(f"cited_only halacha  → skipped         : {n(cited_hal)}")
 
     rows = await pool.fetch(
         "SELECT coalesce(metadata_extraction_status,'NULL') s, count(*) n "
@@ -76,6 +90,13 @@ async def main() -> int:
     print("\nresulting metadata_extraction_status distribution:")
     for r in rows:
         print(f"  {r['s']:<12} {r['n']}")
+    hal_rows = await pool.fetch(
+        "SELECT coalesce(halacha_extraction_status,'NULL') s, count(*) n "
+        "FROM case_law GROUP BY 1 ORDER BY 2 DESC"
+    )
+    print("\nresulting halacha_extraction_status distribution:")
+    for r in hal_rows:
+        print(f"  {r['s']:<12} {r['n']}")
     return 0