fix(halacha): re-extraction preserves chair-approved halachot (INV-G10, #108)

תיקון data-loss: reset_halacha_extraction ביצע DELETE ללא-תנאי לפני חילוץ-מחדש;
קריסה בין המחיקה לאחסון הראשון מחקה את כל אישורי-היו"ר והשאירה את הרשומה תקועה
status='processing' עם 0 שורות (תקרית עמיאל 8126-03-25, 2026-06-08).

עכשיו המחיקה מחריגה review_status IN ('approved','published') — אישור אנושי לא
נמחק בשקט (INV-G10). ה-dedup-on-insert של store_halachot_for_chunk מדלג על חילוץ
טרי שמשכפל מאושרת שנשמרה, כך שאין כפילות. reset מחזיר {deleted, preserved},
וה-extractor מתעד כמה מאושרות נשמרו (provenance, G9).

עמידות מלאה מול מוות-תהליך (OOM) נשארת ל-X16/#114 (durable resume) — זה תנאי-מקדים.

בדיקה: test_halacha_reextract_preserves_approved.py (offline SQL-capture) מאמת
שה-DELETE מחריג approved/published; 64 בדיקות-הלכה קיימות עוברות.

Invariants: G10 (שער-יו"ר — אישור לא נמחק), G1 (תיקון במקור), G9 (provenance).

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

mcp-server/src/legal_mcp/services/db.py

@@ -4157,17 +4157,44 @@ async def store_halachot(case_law_id: UUID, halachot: list[dict]) -> int:
     return len(halachot)
 
 
-async def reset_halacha_extraction(case_law_id: UUID) -> None:
-    """Force a clean re-extraction: wipe halachot + clear per-chunk checkpoints
-    so every chunk is re-processed (used by explicit re-extract, not resume)."""
+async def reset_halacha_extraction(case_law_id: UUID) -> dict:
+    """Prepare a clean re-extraction WITHOUT destroying chair-approved work.
+
+    Deletes only un-reviewed halachot (``review_status NOT IN ('approved',
+    'published')``) and clears per-chunk checkpoints so every chunk is
+    re-processed. Chair-approved / published halachot are PRESERVED — INV-G10:
+    a human approval is never silently deleted by a re-extraction. The
+    re-extractor's dedup-on-insert (:func:`store_halachot_for_chunk`) skips any
+    freshly extracted halacha that duplicates a preserved one, so approvals
+    survive without producing duplicates.
+
+    History: this once wiped ALL halachot first, then re-extracted — a crash
+    between the wipe and the first chunk's store lost every approval and left
+    the row stuck ``status='processing'`` with 0 rows (the 2026-06-08 amiel
+    incident, TaskMaster #108). Durable resume of the whole pipeline is X16/#114.
+
+    Returns ``{"deleted": N, "preserved": M}``.
+    """
     pool = await get_pool()
     async with pool.acquire() as conn:
         async with conn.transaction():
-            await conn.execute("DELETE FROM halachot WHERE case_law_id = $1", case_law_id)
+            preserved = await conn.fetchval(
+                "SELECT COUNT(*) FROM halachot WHERE case_law_id = $1 "
+                "AND review_status IN ('approved', 'published')", case_law_id,
+            )
+            tag = await conn.execute(
+                "DELETE FROM halachot WHERE case_law_id = $1 "
+                "AND review_status NOT IN ('approved', 'published')", case_law_id,
+            )
             await conn.execute(
                 "UPDATE precedent_chunks SET halacha_extracted_at = NULL "
                 "WHERE case_law_id = $1", case_law_id,
             )
+    try:
+        deleted = int(str(tag).split()[-1])
+    except (ValueError, IndexError):
+        deleted = 0
+    return {"deleted": deleted, "preserved": int(preserved or 0)}
 
 
 async def mark_all_chunks_extracted(case_law_id: UUID) -> int: