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_pair_corpus.py

@@ -370,55 +370,6 @@ def test_known_limit_pairs_are_evaluated_but_excluded_from_achievable(tmp_path:
     assert set(result.per_mechanism) == {"cross_artifact", "source_reaches_sink"}
     assert result.loss["sast"]["known_limit"] == 1
     assert "overlay" in result.scorers["sast"] and "inventory" in result.scorers["sast"]
-    assert result.degradations and result.degradations[0]["reason"] == "hunter_model_unavailable"
-
-
-def test_hunter_scorer_credits_a_family_tag_and_never_the_words(tmp_path: Path) -> None:
-    """learning-harness Req 3.1/3.6 replaced this path's rule: the family tag decides, the rationale never does."""
-    from openultrasast.findings import StaticFinding
-
-    src = "from flask import request\n\ndef f():\n    return eval(request.args.get('x'))\n"
-    row = ExpectedFinding(
-        cwe="CWE-95",
-        vulnerability_class="code injection",
-        path="app.py",
-        evidence="",
-        sink="eval",
-        function="f",
-        mechanism="source_reaches_sink",
-    )
-
-    def scripted(tags: list[str]):  # type: ignore[no-untyped-def]
-        def scan(root: Path) -> list[StaticFinding]:
-            if "eval(" not in (root / "app.py").read_text():
-                return []
-            return [
-                StaticFinding(
-                    finding_id="tool-hunter:app.py:4",
-                    path="app.py",
-                    title="eval of request input",
-                    severity="high",
-                    confidence="medium",
-                    evidence_level="suspicion",
-                    rationale="CWE-95 eval executes attacker input",
-                    line=4,
-                    function_name=None,
-                    reachability_status="unknown",
-                    reachability_evidence=[],
-                    reachability_conditions=[],
-                    tags=tags,
-                    ranking_priority=1.0,
-                )
-            ]
-
-        return scan
-
-    case = _case(tmp_path, "hunt", src, "def f():\n    return 1\n", row)
-    tagged = evaluate_catalog([case], hunter=scripted(["family:injection"])).scorers["sast"]["hunter"]
-    assert tagged.pair_correct == 1 and tagged.detected_vuln == 1 and tagged.silent_fix == 1
-    # the same finding, the same words, no family tag: not a detection, because the rationale is not evidence
-    untagged = evaluate_catalog([case], hunter=scripted([])).scorers["sast"]["hunter"]
-    assert untagged.detected_vuln == 0 and untagged.silent_fix == 1
 
 
 def test_juliet_strcpy_pair_bodies_differ_and_java_hash_is_known_limit() -> None: