feat(gate): detect translation leaves stranded on a previous source revision (#12067)

* feat(gate): detect translation leaves stranded on a previous source revision

`os i18n extract --fill=default` fills gaps only, so the ordinary sequence —
extract, revise the source string, extract again — rewrites `en` (never merged,
#8543) and silently keeps the previous source text in every translated locale.
The bundle stays in sync BY KEY, so `check:i18n` reports OK, and the leaf is
still present, so `check:i18n-coverage` counts it translated: it tests presence,
not freshness. Measured on PR #11659 at bbe0b17 — three locales serving a
602-char superseded draft of a 411-char help string under 31 green checks.

"Flag untranslated leaves that no longer match the current source" cannot be
implemented as stated: "untranslated" has exactly one observable spelling today,
EQUAL to the current source, so the two predicates are complementary rather than
conjoinable. Measured on this tree, 2648 of 3010 leaves differ from `en` — that
set is essentially every correct translation in the repo.

What IS decidable without provenance is cross-locale agreement: two different
target languages do not independently produce byte-identical prose, so when they
hold the same bytes neither translated it — both were filled from the source, and
if those bytes are not the CURRENT source they are a previous revision of it.
Three further conditions each close one measured false-positive class (47 -> 11
-> 5); none of them is a threshold.

Existing drift is frozen with a reason per entry, all five of it real and
pre-existing; NEW drift fails. Repairing those bundles is deliberately not this
change — #11671 was re-routed to the tooling lane for that reason.

Needs no workspace build (it reads the committed bundles as text), so it runs in
`Lint & Repo Gates` rather than with its two i18n siblings in `typecheck`.

Part of #11671

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UjM2ia8Av1v5NqfqQEQmC6

* fix(gate): put check-i18n-stale-fill's dispatch behind isEntrypoint

The module exports its rules so they can be imported; without the guard that
import runs the whole gate inside the importer and exits its process — the
silent-success direction scripts/invoked-as.mjs documents. check:entry-guard
caught it.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UjM2ia8Av1v5NqfqQEQmC6

* chore(pm): record check:i18n-stale-fill's bare-root verdict as REFUSE-UNSPELLABLE

The new gate reaches its population through the same findExtractConfigs walk
check:i18n and check:i18n-coverage use, so it joins them rather than being
judged apart: the walk recurses across the root, but what it ADMITS is a
filename-and-segment test — 9 of 5185 tracked files (0.17%).

Measured rather than inherited: both star spellings of the real population
collapse to the same malformed double-separator prefix, and hintCovers matched
it against NOTHING when checked against a real config path; the only spellings
that cover anything collapse to the bare root. So a declaration is either a live
hint over zero files or a hint over 5185 files to reach 9 — and declaring a root
the gate does not read wholesale is the costlier error.

Only the TRIAGE map moves. The recogniser, the sweep and the self-test are
untouched.

Part of #11671

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UjM2ia8Av1v5NqfqQEQmC6

---------

Co-authored-by: Claude <noreply@anthropic.com>

Author: yinlianghui

.github/workflows/lint.yml

@@ -1388,6 +1388,19 @@ jobs:
       - name: Reserved-word ("role") docs ratchet
         run: pnpm check:role-word
 
+      # #11671: `os i18n extract --fill=default` fills GAPS only, so revising a
+      # source string rewrites `en` and strands the previous source text in every
+      # other locale. The bundle stays in sync BY KEY, so `check:i18n` reports OK
+      # and `check:i18n-coverage` counts the leaf translated (it tests presence,
+      # not freshness) — measured on PR #11659, three locales serving a 602-char
+      # superseded draft of a 411-char help string under 31 green checks.
+      # This gate reads the committed bundles as TEXT — no workspace build — which
+      # is why it sits here rather than with its two i18n siblings in `typecheck`.
+      # Existing drift is frozen in scripts/i18n-stale-fill-baseline.json with a
+      # reason per entry; NEW drift fails. Improvements ratchet down via --update.
+      - name: Stale translation fills (i18n) ratchet
+        run: pnpm check:i18n-stale-fill
+
       # #6319: content/docs/getting-started/quick-reference.mdx is the protocol
       # index, and each "## <Name> Protocol (N schemas)" heading is a DECLARATION
       # about the table under it. The page is hand-written (build-docs.ts writes

package.json

@@ -32,6 +32,7 @@
     "i18n:extract": "tsx packages/cli/bin/run-dev.js i18n extract packages/platform-objects/scripts/i18n-extract.config.ts --locales=zh-CN,ja-JP,es-ES --fill=default --out=packages/platform-objects/src/apps/translations",
     "check:i18n": "node scripts/check-i18n-bundles.mjs --self-test && node scripts/check-i18n-bundles.mjs",
     "check:i18n-coverage": "node scripts/check-i18n-coverage.mjs --self-test && node scripts/check-i18n-coverage.mjs",
+    "check:i18n-stale-fill": "node scripts/check-i18n-stale-fill.mjs --self-test && node scripts/check-i18n-stale-fill.mjs",
     "check:app-nav-i18n": "pnpm --filter @objectstack/cli run check:app-nav-i18n",
     "check:nul-bytes": "node scripts/check-nul-bytes.mjs --self-test && node scripts/check-nul-bytes.mjs",
     "check:entry-guard": "node scripts/check-entry-guard.mjs --self-test && node scripts/check-entry-guard.mjs",