docs(scripts): ratchet-remedy corpus — the i18n-walk-parity ledger no longer ships empty (#19154)

Fixes #15189

Clause-②: no

## What this changes

One file, one entry, two prose sentences:
`scripts/check-ratchet-remedy-authority.mjs`'s
hand-classified corpus entry for `check-i18n-walk-parity.mjs` described
that gate's
exemption ledger as shipping **EMPTY**. It has not been empty since the
maintainer ruled,
on the parity gate's own first red, that the unwalked groups enter it
with their reasons —
and the described file records that move itself.

Both sentences now say what is true. Nothing else moved: the
classification, the
`marked`-rather-than-`refused` reason, the ledger, its ceiling and its
`MAINTAINER-ONLY` marking are all untouched, and
`check-i18n-walk-parity.mjs` is not
edited at all.

## The number question, and the choice taken

The new prose carries **no count**. The card notes this family's failure
mode is precisely
a number copied out of another file going stale, and triage reported
this as the seventh
instance. Two further facts made the number-free wording the cheaper one
here:

- the ledger is **shrink-only**, so any count written here is only ever
waiting to be wrong;
- one of its three entries (`settings`) is declared in that file as a
**DEFERRAL** held on
#15178, so the ledger is expected to shrink — a quoted "three" would be
stale by design.

What the corpus entry needs to carry is that the ledger is shrink-only
and that growing it
is the maintainer's act — that, and not its size, is what makes `marked`
the honest verdict
rather than `refused`. Triage's literal wording ("filled to three …
shrink-only from
three") would have been equally true today; the dispatch left the choice
open and this is
the half that does not need re-syncing. ⛔ No house rule is proposed here
— whether prose
may ever quote another file's number is explicitly another card's
question.

## Live reading of the ledger being described

Read in-worktree at **2026-09-19T00:16:13Z** on `origin/main` @
`87f776bd7`, not copied
from the card:

```
scripts/check-i18n-walk-parity.mjs:209   KNOWN_NO_EXTRACTOR_FACE keys: messages, settingsCommon, settings   (3)
scripts/check-i18n-walk-parity.mjs:232   LEDGER_CEILING = 3
```

Same values the dispatch read two minutes earlier; no divergence to
report.

## Evidence

**Both batteries, before and after** — they are two different batteries
and are reported
as two:

| invocation | before | after |
|---|---|---|
| `node scripts/check-ratchet-remedy-authority.mjs --self-test` | exit
**0** | exit **0** |
| `node scripts/check-ratchet-remedy-authority.mjs` (bare) | exit **0**
| exit **0** |

Bare run printed the identical verdict line on both sides — `257 scripts
swept; 15 mark the
expanding remedy, 5 turn it down outright, 237 hand out no
ratchet-expanding remedy.
Control corpus: 30 hand-classified scripts, set-equality audited both
ways.` — so no
verdict moved.

`SELF_TEST_BATTERY_FLOOR` reads **21** before and after. The corpus
entry's battery is
**(16) The corpus-scale positive control**, pinned count **1** before
and after; the
self-test's green is itself the assertion that every declared battery
met its floor.
Nothing in the roster moved.

**Residual reading, with a firing control taken from the probed file
after the diff
existed** (the control was derived by listing terms present in the file
and absent from
`git diff BASE HEAD -- PATH`, not pre-picked):

```
ships EMPTY   -> 0     (also 0 case-insensitively)
ships empty   -> 0
battery       -> 48 lines in the file, 0 in the diff   (this control fires)
```

**The classification survived verbatim:**

- `expect: 'marked'` — 15 occurrences in the file before and after;
**0** of them appear as
  a `+`/`-` line in the diff.
- the `refused`-would-be-untrue comment block (the reason the entry is
`marked` rather than
`refused`) — present once before and after, and it does not appear as a
`+`/`-` line.
- inside the one `why` line that did change, the trailing **174 bytes**
are byte-identical:
`; emitting the group is the author's remedy and is offered first, while
exempting one is
a decision with an owner, so that path is named with its owner rather
than denied.',`
  Only the leading false clause was replaced.

**Gates** derived in-worktree with
`node scripts/pm/dispatch-gates.mjs --commands --repo
objectstack-ai/objectstack` over the
real changed-file set (1 path) — 28 commands; every one that has
reported so far is exit
**0**. By the tool's own warning that list is not a complete account of
CI.

## Changeset

`skip-changeset`. Verified rather than assumed: every published package
declares
`files: ["dist","README.md","CHANGELOG.md"]`, all package-relative, and
the repo-root
package is `private: true` with no `files`. The repo-root `scripts/`
directory is inside no
package and reachable from no published `files[]` entry, so this diff
moves no published
byte.

---
_Generated by [Claude
Code](https://claude.ai/code/session_017ef78bLdybu3AffehKkhfk)_

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

Author: os-try-charles

scripts/check-ratchet-remedy-authority.mjs

@@ -883,14 +883,15 @@ const CONTROL = {
   // (#14653). Same reading as the seventh, eighth and ninth: the gate hands out
   // two remedies with two different owners. Emitting the group in the CLI
   // extractor is the landing author's and touches no ratchet, so it is offered
-  // freely and first. The other path adds a group to a shrink-only exemption
-  // ledger that ships EMPTY by triage ruling — the author excusing themselves
-  // from the parity check they just failed — so it is named with its owner.
+  // freely and first. The other path adds a group to an exemption ledger the
+  // maintainer has since FILLED by ruling on that gate's own first red, and
+  // which only shrinks from there — the author excusing themselves from the
+  // parity check they just failed — so it is named with its owner.
   // `refused` would be untrue: a group keyed by strings no stack config declares
   // genuinely has no extractor face, so there IS a legitimate act here.
   'check-i18n-walk-parity.mjs': {
     expect: 'marked',
-    why: 'PR for #14653. Its exemption ledger ships empty and only shrinks; emitting the group is the author\'s remedy and is offered first, while exempting one is a decision with an owner, so that path is named with its owner rather than denied.',
+    why: 'PR for #14653. Its exemption ledger was filled by maintainer ruling on its own first red and only shrinks from there; emitting the group is the author\'s remedy and is offered first, while exempting one is a decision with an owner, so that path is named with its owner rather than denied.',
   },
 
   'check-adr-links.mjs': {