test(quality): the boundary bar is a round share that keeps every verdict — red

The bar was the known-good arm's own measured fraction, which names the
document it was read off. The test asks for 999/1000 and for the verdict
of every recorded build to stand: the known-good arm passes; an older build
at 99.7 %, the known-bad arm at 41.6 % and five of six fail.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

tests/test_quality.py

@@ -240,12 +240,12 @@ def test_the_no_source_file_row_reports_no_document_count(tmp_path: Path) -> Non
 # (`docs/2026-09-12-g37-terskler.md` SS 7):
 #
 # 1. The fasit's `norm` key is reproduced from its own `title` by stripping all
-#    whitespace and lowercasing -- 2 761 of 2 761 rows, so the normalisation is
+#    whitespace and lowercasing -- N of N rows, so the normalisation is
 #    not a guess.
-# 2. That normalisation ALONE reaches 22 of 2 761 on the known-good arm, not
+# 2. That normalisation ALONE reaches 22 of N on the known-good arm, not
 #    99 %, because okf's default route moves the numbering token STS glues into
 #    `<title>` over into the concept id. The second match form -- the
-#    (directory, residual title) pair -- is what takes it to 2 759 of 2 761.
+#    (directory, residual title) pair -- is what takes it to all but two of N.
 
 
 def _fasit(path: Path, titles: list[str]) -> Path:
@@ -269,7 +269,7 @@ def _declaring(titles: list[str]) -> list[tuple[str, str, str]]:
 def test_the_normalisation_reproduces_the_fasit_key() -> None:
     """`norm` is whitespace-stripped, lowercased `title` -- not a guess.
 
-    Measured over the shipped fasit before anything was written: 2 761 of 2 761
+    Measured over the shipped fasit before anything was written: N of N
     rows reproduce, and the real file pins it below where it exists.
     """
     assert quality.normalise_title("  2.1 Hoved Kapitler\n") == "2.1hovedkapitler"
@@ -304,8 +304,8 @@ def test_a_bundle_recovering_every_declared_boundary_passes(tmp_path: Path) -> N
 def test_a_bundle_recovering_few_declared_boundaries_fails(tmp_path: Path) -> None:
     """The known-negative for the same rule, and the arm it was built for.
 
-    `860019-mdb-100` recovers 1 148 of 2 761. Here five of six is already worse
+    The known-bad HTML arm recovers 41.6 %. Here five of six is already worse
-    than the bar of 2 759/2 761 -- the bar is that tight, which is a property of
+    than the bar of 99.9 % -- the bar is that tight, which is a property of
     a regression bar set at its reference and is said out loud in the document.
     """
     titles = [f"{n} Kapittel {n}" for n in range(1, 7)]
@@ -330,15 +330,15 @@ def test_a_bundle_recovering_few_declared_boundaries_fails(tmp_path: Path) -> No
 def test_the_numbering_token_is_matched_through_the_concept_directory(tmp_path: Path) -> None:
     """The second match form, and the reason P1's literal reading is not enough.
 
-    STS glues the numbering into the title (`11.1Fastmerker`); okf's default
+    STS glues the numbering into the title (`11.1Grunndeiger`); okf's default
     route moves it into the concept id (`11-1/...`) and keeps the residual as
     the title. Measured on the known-good arm: the literal form alone reaches
-    22 of 2 761, the pair form 2 737, either 2 759.
+    22 of N, the pair form N − 24, either N − 2.
     """
-    titles = ["11.1Fastmerker"]
+    titles = ["11.1Grunndeiger"]
     bundle = _bundle(tmp_path / "paired", [("11-1/p1", "doc.pdf", "Body.")])
     (bundle / "11-1/p1.md").write_text(
-        _FRONTMATTER.format(title="Fastmerker", source_file="doc.pdf", body="Body."),
+        _FRONTMATTER.format(title="Grunndeiger", source_file="doc.pdf", body="Body."),
         encoding="utf-8",
         newline="",
     )
@@ -438,6 +438,24 @@ def test_the_boundary_row_says_its_threshold_rests_on_one_corpus(tmp_path: Path)
     assert "N = 1" in rendered
 
 
+def test_the_boundary_bar_is_a_round_share_that_keeps_every_recorded_verdict() -> None:
+    """99.9 %, just under the known-good arm's measured share.
+
+    The bar was the known-good arm's own fraction, which named the document it
+    was read off. A round share just under it gives every build measured the
+    verdict it had: the known-good arm (all but two of its declared sections)
+    passes, an older build of the same product at 99.7 % and the known-bad arm
+    at 41.6 % fail, and so does five of six.
+    """
+    bar = quality.BOUNDARY_THRESHOLD
+    assert bar.as_share() == "999/1000"
+    assert not bar.undercut_by(999, 1000)
+    assert not bar.undercut_by(1998, 2000)
+    assert bar.undercut_by(997, 1000)
+    assert bar.undercut_by(416, 1000)
+    assert bar.undercut_by(5, 6)
+
+
 def test_the_boundary_threshold_names_its_corpus_and_denominator() -> None:
     bar = quality.BOUNDARY_THRESHOLD
     assert bar.limit_declared >= quality.MIN_DECLARED_FOR_A_THRESHOLD