fix(gsm8k): soften microscope live pins and hygiene for PR #806

Remove exact live top-bucket count assertions per #804 doctrine; keep
structural invariants and case 0002 Gate A2a seed pin. Fix EOF newlines
and duplicate _PRIMITIVE_BY_SUBFAMILY key.

docs/analysis/core-problem-solving-capability-roadmap-v2-2026-06-17.md

@@ -80,4 +80,4 @@ Additive comparative is **not** roadmap Gate A2. Roadmap Gate A2 remains partiti
 
 1. **Ratify** Gate A2a unit partition / chunking primitive (shared narrow primitive — see lookback evidence). Implementation follows ratification only.
 2. Preserve wrong=0 and monotonic counts on every future slice.
-3. Gate A1b / Comparative-A2 (additive comparative) and Inc4 denom-state remain deferred until separately ratified.
+3. Gate A1b / Comparative-A2 (additive comparative) and Inc4 denom-state remain deferred until separately ratified.

scripts/gsm8k_post_gate_a1_frontier_microscope.py

@@ -127,7 +127,6 @@ _PRIMITIVE_BY_SUBFAMILY: dict[str, str] = {
     "numeric_expression_duration": "duration_multiplier",
     "numeric_expression_recurrence": "recurrence_frame",
     "partition_statement_parser_gap": "unit_partition",
-    "affine_equation_fraction_target": "fractional_delta",
     "unclassified": "diagnostic_hold",
     "inverse_residual_more": "inverse_residual_question",
     "ratio_times_entity": "ratio_question_frame",

tests/test_gsm8k_frontier_report.py

@@ -215,8 +215,8 @@ def test_markdown_render_is_stable_and_mentions_rate():
     assert render_markdown(summary) == md
 
 
-def test_frontier_report_aligns_with_post_gate_a1_microscope_top_buckets():
-    """Live ephemeral top buckets match the post-Gate-A1 microscope partition."""
+def test_frontier_report_aligns_with_post_gate_a1_microscope_structure():
+    """Live microscope stays structurally complete with closed injector buckets."""
     from scripts.gsm8k_post_gate_a1_frontier_microscope import build_microscope_report
 
     cases_path = _REPO_ROOT / "evals/gsm8k_math/train_sample/v1/cases.jsonl"
@@ -227,7 +227,9 @@ def test_frontier_report_aligns_with_post_gate_a1_microscope_top_buckets():
     ]
     microscope = build_microscope_report(cases)
     assert_monotonic_serving_counts(microscope["counts"])
-    assert microscope["top_buckets"]["recognized_no_injection"] == 31
+    refused = microscope["counts"]["refused"]
+    assert len(microscope["refusal_table"]) == refused
+    assert sum(microscope["top_buckets"].values()) == refused
     assert microscope["closed_injector_buckets"]["rate_with_currency_no_injection"] == 0
     assert microscope["closed_injector_buckets"]["comparative_with_unit_no_injection"] == 0
 

tests/test_gsm8k_post_gate_a1_frontier_microscope.py

@@ -15,6 +15,29 @@ from tests.gsm8k_train_sample_baseline import assert_monotonic_serving_counts
 _REPO_ROOT = Path(__file__).resolve().parents[1]
 _CASES_PATH = _REPO_ROOT / "evals/gsm8k_math/train_sample/v1/cases.jsonl"
 
+_APPROVED_TOP_BUCKETS = frozenset({
+    "recognized_no_injection",
+    "no_admissible_statement",
+    "no_admissible_question",
+    "no_solvable_branch",
+    "incomplete_reading",
+    "other_refused",
+    "other",
+})
+
+_REFUSAL_TABLE_FIELDS = frozenset({
+    "case_id",
+    "verdict",
+    "reason",
+    "top_refusal_bucket",
+    "subfamily",
+    "matched_recognizer_category",
+    "first_blocking_layer",
+    "candidate_next_primitive",
+    "expected_movement",
+    "evidence_snippet",
+})
+
 
 def _load_cases() -> list[dict]:
     return [
@@ -36,43 +59,26 @@ def test_live_microscope_meets_monotonic_contract_and_closed_injectors():
 def test_live_microscope_refusal_partition_is_complete():
     summary = build_microscope_report(_load_cases())
 
-    assert summary["counts"]["refused"] == 44
-    assert len(summary["per_case"]) == 44
-    assert len(summary["refusal_table"]) == 44
-    assert sum(summary["top_buckets"].values()) == 44
+    refused = summary["counts"]["refused"]
+    assert len(summary["refusal_table"]) == refused
+    assert len(summary["per_case"]) == refused
+    assert sum(summary["top_buckets"].values()) == refused
+    assert set(summary["top_buckets"]) <= _APPROVED_TOP_BUCKETS
+    assert summary["top_buckets"].get("recognized_no_injection", 0) >= 0
 
-    required = {
-        "case_id",
-        "verdict",
-        "reason",
-        "top_refusal_bucket",
-        "subfamily",
-        "matched_recognizer_category",
-        "first_blocking_layer",
-        "candidate_next_primitive",
-        "expected_movement",
-        "evidence_snippet",
-    }
     for row in summary["refusal_table"]:
-        assert required <= set(row.keys())
-
-    # Post-Gate-A1 stable top-level buckets (ephemeral main @ bb083004 family)
-    assert summary["top_buckets"]["recognized_no_injection"] == 31
-    assert summary["top_buckets"]["no_admissible_statement"] == 7
-    assert summary["top_buckets"]["no_admissible_question"] == 5
-    assert summary["top_buckets"]["no_solvable_branch"] == 1
+        assert _REFUSAL_TABLE_FIELDS <= set(row.keys())
+        assert row["verdict"] == "refused"
+        assert row["reason"]
+        assert row["evidence_snippet"]
 
 
-def test_live_microscope_dcs_and_slice_candidate_pins():
+def test_live_microscope_partition_seed_case_is_tagged():
     summary = build_microscope_report(_load_cases())
-
-    assert summary["recognized_no_injection_by_category"]["discrete_count_statement"] == 19
-    assert summary["dcs_no_injection_subfamilies"]["dcs_misroute_unit_partition"] == 1
     assert (
         "gsm8k-train-sample-v1-0002"
         in summary["implementation_slice_candidates"]["partition_chunking"]["case_ids"]
     )
-    assert summary["no_solvable_branch_subfamilies"]["rate_graph_unsolvable"] == 1
 
 
 def test_microscope_output_is_deterministic():
@@ -112,4 +118,4 @@ def test_case_0002_ratification_candidate_fields():
     assert (
         summary["recommended_next_ratification_candidate"]
         == "Gate A2a unit_partition / chunking primitive"
-    )
+    )