feat(learning): chair-feedback style corrections auto-flow to the writer (P1 #6)

The chair-feedback chain was DEAD: 27 feedback rows captured, 0 ever became a
lesson the writer reads (no weekly-analysis job ever existed). The chair's own
corrections — the most authoritative signal of all, and exactly "learn from every
returned draft" — went nowhere.

decision_lessons can't carry them: it's FK-coupled to a style_corpus row (a signed
final), which a case-in-progress doesn't have. So chair STYLE feedback instead
rides the discussion_rules channel that already reaches the writer for all blocks —
the same path /training promote uses.

- db.append_global_rule: the locked read-modify-write append, extracted from web
  `_append_methodology_override` into one shared impl (G2). The web function is now
  a thin wrapper that seeds defaults; chair-feedback calls it directly.
- record_chair_feedback (MCP tool): a STYLE-category correction (style/wrong_tone/
  wrong_structure) with a lesson_extracted flows immediately to discussion_rules.
  The chair IS the gate — no separate approval (INV-LRN1 graduated gate). SUBSTANCE
  feedback (missing_content/factual_error/other) is case-specific → recorded only.

Invariants: INV-LRN1 (chair-authored style = highest authority, flows; substance
not auto-flowed), G2 (single append impl shared by promote + feedback), INV-LRN5
(style channel only).

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

mcp-server/src/legal_mcp/tools/workflow.py

@@ -394,13 +394,35 @@ async def record_chair_feedback(
         lesson_extracted=lesson_extracted,
     )
 
+    # Auto-flow chair-authored STYLE feedback to the writer (closes the dead
+    # chair_feedback→lesson chain — 27 feedback rows had produced 0 lessons). The
+    # chair is the highest authority, so a style correction she writes flows
+    # immediately — the chair IS the gate. It rides the SAME discussion_rules channel
+    # promote uses (db.append_global_rule, G2), reaching every block, without the
+    # style_corpus coupling decision_lessons require. SUBSTANCE feedback
+    # (missing_content/factual_error/other) is case-specific → recorded only.
+    # (INV-LRN1 graduated gate; 07-learning §1.2.)
+    _STYLE_FB = {"style", "wrong_tone", "wrong_structure"}
+    flowed = 0
+    if lesson_extracted.strip() and category in _STYLE_FB:
+        try:
+            flowed = await db.append_global_rule(
+                "discussion_rules", "universal", [lesson_extracted.strip()],
+            )
+        except Exception as e:
+            logger.warning("chair-feedback auto-flow failed for %s: %s", case_number, e)
+
+    msg = f"הערה נרשמה בהצלחה. קטגוריה: {category}."
+    if flowed:
+        msg += " הלקח (סגנון) זרם אוטומטית לכותב."
     return ok({
         "feedback_id": str(feedback_id),
+        "flowed_to_writer": bool(flowed),
         "next_steps": [
             "כדי להפיק לקח מההערה, הפעל: analyze_chair_feedback",
             "כדי לסמן כמטופל: resolve_chair_feedback",
         ],
-    }, message=f"הערה נרשמה בהצלחה. קטגוריה: {category}.")
+    }, message=msg)
 
 
 _CURATOR_FINDING_CATEGORIES = {"style", "structure", "lexicon", "tabular", "general"}