fix(dedup): a rule or preference create surfaces what it overlaps by meaning (#4134)

find_duplicate_rule was title-only, on the stated premise that rules are not
a semantic-retrieval surface - false since rules were embedded. A preference
restating a rule under another title passed untouched, and since both kinds
share one ranking, the weaker label could arrive alone.

find_overlapping_rules queries semantic_search_rules with the rule_document
shape, both kinds, in the scope the new record ranks in (global: every rule
the caller owns; project: global + that project). All three MCP create doors
call it before creating and return overlaps + overlap_note on the reply.

It advises rather than blocks, on measurement: across 16 sampled records the
nearest DISTINCT neighbour reached 0.853, while a true rewording scored 0.850.
No threshold separates the bands, so the floor (0.80) sits below the
restatement and the author judges. The stale docstring is corrected.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01821k5B3Ysecp9fNYs92Kuy

tests/conftest.py

@@ -147,3 +147,20 @@ def _no_rule_arm():
     with patch("scribe.services.plugin_context.semantic_search_rules",
                AsyncMock(return_value=[])):
         yield
+
+
+@pytest.fixture(autouse=True)
+def _no_rule_overlap():
+    """Stub the rule/preference create path's overlap check (#4134).
+
+    The same reason as _no_rule_arm, one door over: every create_rule /
+    create_project_rule / create_preference now asks semantic_search_rules
+    whether an existing record answers the same moment, so each existing
+    rule-tool unit test would load the embedding model through a call it never
+    meant to make. The check's own behaviour is tested in
+    tests/test_rule_overlap_gate.py, which binds the real function at import
+    time — before this patch runs — and stubs the search beneath it instead.
+    """
+    with patch("scribe.services.dedup.find_overlapping_rules",
+               AsyncMock(return_value=[])):
+        yield