fix(halacha): split authority (derived) from rule_role — stop source-conflation (INV-DM7)

The extractor classified rule_type by SOURCE bindingness (higher-court→binding,
committee→persuasive) instead of by rule KIND. The gold-set proved it: 'binding'
appeared on 19/19 external rulings & 0 committees; 'persuasive' on 13/13
committees & 0 external — only 58% agreement with the human role tags. The two
axes (authority vs rule role) were crammed into one enum.

This splits them per INV-DM7:
- authority (binding/persuasive) — DERIVED from case_law.precedent_level
  (עליון/מנהלי→binding, ועדת_ערר_מחוזית→persuasive), never stored, never
  LLM-guessed. New helper halacha_quality.derive_authority; surfaced read-only
  in list_halachot / goldset_list / search results.
- rule_type — now the rule ROLE only: holding/interpretive/procedural/
  application/obiter. Both extractor prompts unified to this vocabulary;
  _coerce_halacha no longer defaults rule_type from the source; legacy
  binding→holding / persuasive→interpretive fold for safety.

UI: authority shown as a separate read-only badge (gold=מחייב / muted=משכנע)
across the review queue, precedent detail, and gold-set; the gold-set role
selector drops binding/persuasive and adds מהותי (holding).

Migration: scripts/halacha_rule_role_backfill.py re-classifies the 276 pre-split
binding/persuasive rows into a genuine role via local claude_session (run after
deploy). Gold-set correct_type/ai_correct_type 'binding'→'holding' via SQL.

Sources (≥3, per research-decision policy): OASIS LegalRuleML v1.0
(appliesAuthority/Strength as metadata orthogonal to rule logic) · SemEval-2023
Task 6 LegalEval (rhetorical roles by function, authority kept separate) ·
Bluebook signals (weight-of-authority is a separate dimension).

Invariants: ESTABLISHES INV-DM7. Upholds G1 (normalize at source — extractor
classifies role, system derives authority) and G2 (single source of truth —
authority derived, not a parallel stored field). Tests: 211 pass + new
derive_authority/coerce coverage. web-ui build + tsc clean.

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

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

@@ -664,8 +664,10 @@ CREATE TABLE IF NOT EXISTS halachot (
     case_law_id UUID REFERENCES case_law(id) ON DELETE CASCADE,
     halacha_index INTEGER NOT NULL,
     rule_statement TEXT NOT NULL,
-    rule_type TEXT DEFAULT 'binding',
-        -- binding | interpretive | procedural | obiter
+    rule_type TEXT DEFAULT 'interpretive',
+        -- rule ROLE only (INV-DM7): holding | interpretive | procedural |
+        -- application | obiter. authority (binding/persuasive) is DERIVED
+        -- from case_law.precedent_level, never stored here.
     reasoning_summary TEXT DEFAULT '',
     supporting_quote TEXT NOT NULL,
     page_reference TEXT DEFAULT '',
@@ -4052,7 +4054,7 @@ async def store_halachot(case_law_id: UUID, halachot: list[dict]) -> int:
                 case_law_id,
                 i,
                 h["rule_statement"],
-                h.get("rule_type", "binding"),
+                h.get("rule_type", "interpretive"),
                 h.get("reasoning_summary", ""),
                 h["supporting_quote"],
                 h.get("page_reference", ""),
@@ -4193,7 +4195,7 @@ async def store_halachot_for_chunk(
                        VALUES ($1, $2, $3, $4, $5, $6, $7, $8, $9, $10, $11,
                                $12, $13, $14, $15, $16, {reviewed_at_clause})""",
                     case_law_id, base + inserted, h["rule_statement"],
-                    h.get("rule_type", "binding"), h.get("reasoning_summary", ""),
+                    h.get("rule_type", "interpretive"), h.get("reasoning_summary", ""),
                     h["supporting_quote"], h.get("page_reference", ""),
                     h.get("practice_areas", []), h.get("subject_tags", []),
                     h.get("cites", []), confidence, h.get("quote_verified", False),
@@ -4299,6 +4301,8 @@ async def list_halachot(
         d = dict(r)
         if d.get("decision_date") is not None:
             d["decision_date"] = d["decision_date"].isoformat()
+        # authority is DERIVED from the source, never stored (INV-DM7)
+        d["authority"] = halacha_quality.derive_authority(d.get("precedent_level"))
         out.append(d)
     if cluster and out:
         await _annotate_clusters(pool, out)
@@ -4721,7 +4725,7 @@ async def goldset_list(batch: str = "default") -> list[dict]:
         "       g.ai_is_holding, g.ai_correct_type, g.ai_rationale, g.ai_generated_at, "
         "       h.rule_statement, h.supporting_quote, h.reasoning_summary, "
         "       h.rule_type, h.confidence, h.quality_flags, h.review_status, "
-        "       cl.case_number, cl.case_name, cl.source_type "
+        "       cl.case_number, cl.case_name, cl.source_type, cl.precedent_level "
         "FROM halacha_goldset g JOIN halachot h ON h.id = g.halacha_id "
         "LEFT JOIN case_law cl ON cl.id = h.case_law_id "
         "WHERE g.batch = $1 ORDER BY g.created_at, g.id", batch,
@@ -4735,6 +4739,8 @@ async def goldset_list(batch: str = "default") -> list[dict]:
             d["ai_generated_at"] = d["ai_generated_at"].isoformat()
         if d.get("confidence") is not None:
             d["confidence"] = float(d["confidence"])
+        # authority is DERIVED from the source, never stored (INV-DM7)
+        d["authority"] = halacha_quality.derive_authority(d.get("precedent_level"))
         out.append(d)
     return out
 
@@ -4792,7 +4798,7 @@ async def goldset_score(batch: str = "default") -> dict:
     for r in labeled:
         rule = r.get("rule_statement") or ""
         quote = r.get("supporting_quote") or ""
-        rtype = r.get("rule_type") or "binding"
+        rtype = r.get("rule_type") or "interpretive"
         qc = r["quote_complete"] if r["quote_complete"] is not None else True
         truly_bad = r["is_holding"] is False
         flags = halacha_quality.compute_quality_flags(rule, quote, "", qc, rtype)
@@ -4990,6 +4996,8 @@ async def search_precedent_library_semantic(
             _conf = float(d.get("confidence") or 0.0)
             d["score"] = float(d["score"]) + max(_conf * 0.06, 0.0)
             d["type"] = "halacha"
+            # authority is DERIVED from the source, never stored (INV-DM7)
+            d["authority"] = halacha_quality.derive_authority(d.get("precedent_level"))
             results.append(d)
 
     rows = await pool.fetch(chunk_sql, *c_params)