docs(spec): the unresolvable-column entry states MySQL's reach after #8926 — envelope and recoveries both (#9060)

[os-zhuang]

Fixes#9060

Text amendment to an already-registered ADR-0087 entry. The entry id, surface
and replacement prescription are unchanged and no accept/reject behaviour
moves — what changes is the reason, which is upgrader-facing documentation
(the data source for objectstack migrate meta, spec-changes.json and the
generated upgrade guide).

Premise verified on the merged tree first

The card claims the entry's "Reach, stated rather than assumed" paragraph became
false when #8926 merged. Verified against origin/main at 24206416a rather
than taken from the card:

• packages/drivers/driver-sql/src/sql-driver.ts — isUnresolvableColumnError
  now carries a third arm, message.includes("Unknown column '"), alongside the
  SQLite and Postgres arms. Its docblock states the reach as all three dialects
  "since driver-sql: MySQL's unresolvable-column wording is matched by neither arm of the recovery/refusal predicate, so MySQL gets no ADR-0112 envelope and no #3821 recovery #8926".
• sql-driver-unresolvable-where-column-refusal.test.ts — the message-shape
  sweep flips the MySQL WHERE case from recognised: false to true, adds the
  'field list' and 'order clause' spellings, and carries an explicit
  direction pin: "mysql: all three clause positions ride ONE arm — envelope and
  recoveries arrive together (driver-sql: MySQL's unresolvable-column wording is matched by neither arm of the recovery/refusal predicate, so MySQL gets no ADR-0112 envelope and no #3821 recovery #8926)".

So the triage re-pricing holds as written: #9061 delivered full parity, not
the envelope alone. Both halves are stated in the addendum.

What this PR does

Appends a dated addendum to the entry's reason. The historical paragraph is
kept verbatim as the state at registration — it is not rewritten — and the
addendum amends it, stating both halves of what one shared predicate did on
MySQL:

1. The envelope — an unresolvable WHERE column refuses with the same
   INVALID_FILTER / 400 naming the column, instead of travelling out as the
   raw ER_BAD_FIELD_ERROR with the statement's bound literals inlined.
2. The recoveries — MySQL also gained the fix(sharing): 共享规则新建页 — 自定义 widget 未国际化，且「接收方」永远无可选项 #3821 projection and ORDER-BY
   recoveries it never had, so those clause positions now return recovered rows
   where they used to throw.

Both arrive together because ER_BAD_FIELD_ERROR spells every clause position
with one sentence, so all three ride one arm of the predicate — cited to the
widened sweep above. The addendum also records what the ruling did not
change: a dotted filter key is still classified per dialect (Postgres raises
undefined_table, which neither arm matches), the axis #8371 owns.

registry.ts regenerated with gen:migration-registry — never hand-edited.

Measured: two of the four assumed artifacts do not move, and that is correct

The dispatch assumed the chain entry edit ⇒ gen:migration-registry ⇒
gen:spec-changes ⇒ gen:upgrade-guide. Only the first link produces a diff
here. Both generators were run anyway and rewrote their files byte-identically:

• PROTOCOL_VERSION is 17.0.0, so spec-changes.json folds the floor-10 to
  major-17 range (measured: the majors present in the committed manifest are
  11-17).
• This entry is registered under major 18, so it is not projected into
  spec-changes.json or docs/protocol-upgrade-guide.md yet.

check:spec-changes and check:upgrade-guide both pass ("up to date"), which
is synchrony holding rather than a skipped step. Nothing to hand-fix; recorded
so the empty diff is not read as a missed regeneration.

Verification — union run at ad1f96e88 (the final commit)

Build closure first: pnpm --filter '@objectstack/spec^...' build matches no
projects (spec has no workspace deps), so pnpm --filter @objectstack/spec build
was run instead; the working tree stayed clean afterwards, so gen:schema moved
no anchor.

gate	result
spec test	406 files / 10785 tests passed
spec typecheck	pass (tsc --noEmit, scripts project, test-typecheck)
check:migration-registry	registry.ts is current (100 semantic, 30 retired-key, 54 retired-def)
check:spec-changes	spec-changes.json is up to date.
check:upgrade-guide	protocol-upgrade-guide.md is up to date.
check:generated	All 13 generated artifacts are up to date.
check:nul-bytes	OK, 5990 files, no raw control bytes
check-adr-0087-registration	self-test 206 assertions; this PR adds no declared-breaking changeset
check:changeset-gate-self-tests	pass
check:merge-driver	pass
check:spec-parsed-alias	pass
check:type-source-resolution	pass
check:cross-package-test-inputs	pass
check:doc-formula-expressions	pass

Gates beyond the dispatch list were re-derived from the actual changed paths with
node scripts/pm/dispatch-gates.mjs, which added the last eight rows above.

Two of them failed first on an unbuilt worktree, not on this diff:
check:doc-formula-expressions could not resolve
@objectstack/formula/dist/index.mjs, and check-dev-prereqs reports 66 of 67
packages have no dist/. Building @objectstack/formula turned the first green
(24 self-test cases, 22 formula examples, 9 spec TSDoc examples). check-dev-prereqs
still reports the unbuilt workspace — it is a whole-workspace precondition
("run pnpm build"), unrelated to this change and not something a text amendment
can satisfy locally.

Changeset: patch on @objectstack/spec, non-breaking, so no ADR-0087
disposition marker is required (the gate confirms).

Generated by Claude Code

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

This PR changes 1 package(s): @objectstack/spec.

106 hand-written doc(s) reference the affected code — list omitted above 15 rows. Re-derive: node scripts/docs-audit/affected-docs.mjs --json origin/main.

⛔ 7 release-owned page(s) also affected — read-only, see AGENTS.md Documentation Guardrails.

Advisory only. To re-verify, run the docs-accuracy-audit workflow scoped to these files:
node scripts/docs-audit/affected-docs.mjs origin/main → pass the list as args.docs.