feat(model-grounded-detection): group 1, subtract the noise architecture

99 source modules -> 81. 832 tests -> 646. The whole learning/ package (16
modules, ~4,100 lines) and six semantic/ modules removed, with the three gates
byte-identical to 2026-09-06-gate-baseline.txt at every step.

The sequence is what made it safe: vocabularies ported first (1.1), importers
detached second (1.2-1.3), deletion third (1.4). Every import break during the
deletion was therefore a detachment miss in the importer and was fixed there --
no module was restored to satisfy one. Both misses were the same shape and were
fixed as a class: operations.py and shapes.py each imported trailing_name from
the deleted variants.py, so it moved to semantic/ir.py, its natural home, and
both importers changed in one edit.

Ported before deleting: the closed guard vocabulary (verbatim, nothing else
holds it) into model/specs.py, which derives TaintSpec and DominanceSpec from
the retained ruleset facts rather than copying them. Re-homed: families ->
model/taxonomy, candidates -> model/candidates, endpoint -> model/endpoint,
plus Region, labeled_family and labeled_spans. labeled_family no longer falls
back to the LLM classifier -- an unlabeled pair is honestly unknown.

Four deviations, each recorded in the spec notes with its reason. improve/
evolve.py and validator.py were stripped, not deleted: each mixes the mechanism
lever (gone) with the deterministic rule-status loop `ousast improve` still
runs, which has no LLM in it; a test now asserts the lever's absence by symbol
so it cannot creep back. The learning CLI group became `model candidates`,
because deleting it outright would have orphaned the retained enumerator by
removing its only entry point. thinking moved from [learning] to [models].

A gap named rather than papered over: no retained sink fact routes by CWE to
path, output_encoding, untrusted_destination or prototype -- the same hole the
entailment ceiling diagnosed. The test asserts the seeded set and the gap set
exactly, so task 4.2 must update it when it authors those sinks.

The audit found exactly one genuine orphan: vectorstore.py, a bakeoff harness
with no production caller and no CLI, alive only through its own test. The five
named subsystems each carry an explicit decision. Manifest committed, with a
test asserting the tree still matches it.

Group 2 (the Joern go/no-go) has not started.

Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
Claude-Session: https://claude.ai/code/session_01T7XD7x5KonMdmGRkp63EQG

tests/test_semantic_pipeline.py

@@ -1,4 +1,4 @@
-"""MAP overlay, prove filter, reports, pair scoring, mechanism memory."""
+"""MAP overlay, prove filter, reports, pair scoring."""
 
 from __future__ import annotations
 
@@ -15,7 +15,7 @@ from openultrasast.gate import LANGUAGE_MANIFESTS
 from openultrasast.pairs import PairCase, evaluate_pair, load_pair_catalog
 from openultrasast.preprocess import preprocess_repository
 from openultrasast.rank import rank_targets
-from openultrasast.semantic import MechanismStore, OverlayRecord, append_mechanism, filter_promoted_hotspots, order_promotions
+from openultrasast.semantic import OverlayRecord, filter_promoted_hotspots
 from openultrasast.semantic.prove_filter import promoted_findings
 
 
@@ -265,79 +265,3 @@ def test_markdown_report_includes_four_overlay_labels(tmp_path: Path) -> None:
     assert "coverage" in text
 
 
-def test_mechanism_order_boosts_paraphrased_match(tmp_path: Path) -> None:
-    store = MechanismStore(tmp_path / "mechanisms.jsonl")
-    append_mechanism(
-        store,
-        summary="request parameter reaches eval",
-        cwe="CWE-95",
-        language="python",
-        tags=["syscall_entry"],
-        what_made_it_exploitable="dynamic execution of attacker input",
-    )
-    eval_id = "python-unsafe-eval:app.py:1"
-    hash_id = "python-weak-hash:hash.py:1"
-    hotspots = (_hotspot(hash_id, score=5.0, path="hash.py"), _hotspot(eval_id, score=5.0, path="app.py"))
-    records = (
-        _record(hash_id, "promote", path="hash.py", cwe="CWE-327", sinks=("hashlib",), language="python"),
-        _record(eval_id, "promote", sources=("request",), cwe="CWE-95", sinks=("eval",), language="python"),
-    )
-    findings = [_finding(hash_id, path="hash.py"), _finding(eval_id)]
-
-    class _Client:
-        def embed(self, *, model: str, inputs: list[str], timeout_seconds: int = 60) -> list[list[float]]:
-            del model, timeout_seconds
-            vectors: list[list[float]] = []
-            for text in inputs:
-                if "eval" in text.lower() or "CWE-95" in text:
-                    vectors.append([1.0, 0.0])
-                else:
-                    vectors.append([0.0, 1.0])
-            return vectors
-
-    ordered, degradation = order_promotions(hotspots, records, findings, store=store, client=_Client(), model="test-embed")
-    assert degradation is None
-    assert ordered[0].inventory_finding_ids == (eval_id,)
-    assert {item.inventory_finding_ids[0] for item in ordered} == {eval_id, hash_id}
-
-
-def test_mechanism_degrades_to_heuristic_without_key(tmp_path: Path, monkeypatch: pytest.MonkeyPatch) -> None:
-    monkeypatch.delenv("OPENROUTER_API_KEY", raising=False)
-    store = MechanismStore(tmp_path / "mechanisms.jsonl")
-    append_mechanism(
-        store,
-        summary="request parameter reaches eval",
-        cwe="CWE-95",
-        language="python",
-        tags=["syscall_entry"],
-        what_made_it_exploitable="dynamic execution of attacker input",
-    )
-    eval_id = "python-unsafe-eval:app.py:1"
-    hash_id = "python-weak-hash:hash.py:1"
-    hotspots = (_hotspot(hash_id, score=6.0, path="hash.py"), _hotspot(eval_id, score=3.0, path="app.py"))
-    records = (
-        _record(hash_id, "promote", path="hash.py", cwe="CWE-327", sinks=("hashlib",)),
-        _record(eval_id, "promote", sources=("request",)),
-    )
-    findings = [_finding(hash_id, path="hash.py"), _finding(eval_id)]
-    ordered, degradation = order_promotions(hotspots, records, findings, store=store, client=None, model=None)
-    assert degradation == "embeddings_unavailable"
-    assert [item.inventory_finding_ids[0] for item in ordered] == [hash_id, eval_id]
-
-
-def test_mechanisms_are_not_written_from_unadjudicated(tmp_path: Path) -> None:
-    store = MechanismStore(tmp_path / "mechanisms.jsonl")
-    record = OverlayRecord(
-        proposal_id="python-unsafe-eval:app.py:1",
-        path="app.py",
-        line=1,
-        disposition="unadjudicated",
-        reason="parse_failed",
-        cwe="CWE-95",
-        sources=(),
-        sinks=(),
-        sanitizers=(),
-        evidence_level="static_corroboration",
-    )
-    assert record.disposition != "promote"
-    assert store.load() == ()