Thirty fields published a zero for a comparison that never happened

[ChelseaKR]

The defect

FieldSpec.outside_domain answers False for every cell of a field the layer publishes
no coded-value domain for. field_coverage summed that answer for every field alike, so
thirty of the fifty-four measured fields published "outside_published_domain": 0,
with a distinct count of 0 and an empty value object beside it.

That zero is a measurement nobody took. It is indistinguishable in the artifact from
CAUSE, which does have a published domain and does hold every value in it, and it
says the file and the domain agree about a domain that does not exist. Both publishers
state which of their fields they constrain; FIRE_NAME, UNIT_ID, COUNTY,
SITEADDRESS and twenty-six others are free text.

It is live in the published artifacts today, on both pages.

The fact was already known one layer down.
tests/test_schema.py::test_free_text_fields_have_no_domain_to_be_outside_of asserts it
at the FieldSpec level. Nothing carried it into the counting and nothing tested what the
count published, so the artifact contradicted a test in the same suite.

The fix

• The four out-of-domain measures are null together for a field with no published
  domain, the way present_tenths_pct is already null where there is no denominator.
  Zero keeps the narrow meaning it should always have had: the domain was published, it
  was read, and the file holds nothing outside it.
• Every free-text row on both pages now says so: "The layer publishes no coded-value
  domain for this field, so no value here is counted as outside one." The paragraph under
  each field table names all three states a row can be in, so a row's silence means
  exactly one of them.
• outside_published_domain_values is capped at twelve and never said so. No field reaches
  the cap today, which is why nothing has been misread yet; a truncated list that does not
  declare itself reads as the whole set. outside_published_domain_values_listed is
  published beside outside_published_domain_distinct.
• docs/adr/0010, README.md and CHANGELOG.md record the decision and what changed.

Tests

Three new count-level tests in tests/test_measurements.py, two artifact tests over both
published payloads in tests/test_artifacts_and_pages.py, three page tests in
tests/test_pages_html.py including one asserting the overview carries neither note, so
the other two are not passing over every page. Every new test was run against the previous
code and fails there.

site/ and make site-check

site/ is committed, is what the Pages workflow uploads, and is rebuilt here from the
acquired files. tests/test_published_site_is_current.py compares the artifacts' shape, so
the new key would have failed the build until site/ was rebuilt. It compares pages only
outside <main>, so the thirty new page notes would have gone unnoticed.

make site-check closes that half: a rebuild from data/raw/ into build/site-current,
diffed against site/, never writing into site/, and refusing outright when the acquired
files are absent, because a check that could not run is not a check that passed. Per
ADR-0004 it was run against both inputs it exists to catch, a drifted page and a missing
source file, and fails on both. It cannot run in CI: data/raw/ is not in git and never
will be.

Gate

make verify is green locally end to end: lock-check, sync, lint, format, mypy --strict,
682 tests with the coverage floor, pip-audit, html-validate, axe in jsdom and in Chromium,
Reflow at 320px, npm audit on both trees, and the determinism check. make site-check is
green against the acquired FRAP and DINS files.

🤖 Generated with Claude Code