chore(sync): reconcile the cf5f15e3 and f6d188bd entries with #432

.sync/nuxt-ui.json

@@ -1525,14 +1525,14 @@
       "summary": "fix(Slider): bind form aria attributes on thumbs instead of root (nuxt/ui #6768) — PORT, verbatim, against Range. useFormField's ariaAttrs (aria-invalid, aria-describedby) were spread onto SliderRoot, which renders no role, so assistive technology read an invalid slider as valid and never reached the error text; the widget is the thumb, which carries role=slider, so the attributes move there — both thumb sites, tooltip-wrapped and bare. This fork's component is Range: upstream's Slider was renamed here, recorded in §1 since #423, which exists because THIS commit was first reported as a no-op on the grounds that no Slider component exists. Only the wrapper is renamed — every reka-ui export inside the file keeps its own name, so SliderRoot/SliderThumb needed no rewriting and the diff is upstream's line for line apart from ui -> b24ui. The defect was ours in full: Range.vue:120 carried the identical v-bind=\"{ ...rootProps, ...ariaAttrs }\" and both SliderThumb sites lacked the attributes. Test is upstream's with the guard renamed to name === 'Range'; our FormField.spec.ts keys describe.each off __name, so the block is reachable only if that value is exactly Range — verified by seeing it in the reporter output rather than assumed, since a guard that never matches is a test that passes without running. Two mutations, each failing a different assertion: full revert fails the role check (expected undefined to be 'slider'), half revert (attributes on root AND thumbs) fails the count (length of 1 but got 2). No snapshot moved, and that is the opposite of the usual meaning: Range.spec.ts renders standalone, outside a FormField, so ariaAttrs is empty and moving an empty object changes no rendered attribute — the 20 role=slider occurrences there come from reka-ui and never carried aria-invalid. The snapshots could not have caught this defect, which is why the fix is pinned by a FormField-level test. lint, typecheck, test (6708 passed, 6 skipped, 294 files), build green; no docs:generate, no docs/ touched."
     },
     "cf5f15e3c4d35e083003fc89700d063623635dbf": {
-      "pr": "pending-merge",
-      "b24ui_sha": "pending-merge",
+      "pr": 432,
+      "b24ui_sha": "807d52efaf8a29cc23b97df3a047c082752bfad0",
       "decision": "no-op",
       "summary": "docs(showcase): add codemo.dev (nuxt/ui #6663) — NO-OP: two lines in docs/content/showcase.yml plus a screenshot asset, adding Codemo to the list of sites built with Nuxt UI. We have that file, but it lists projects built with Bitrix24 UI and its own description says so, so adding a nuxt/ui site would make the page assert something false. Content, not structure — nothing to port and nothing to adapt. Recorded so the next porter does not re-derive it."
     },
     "f6d188bd785d40f2a2d721e3400091cc19cee236": {
-      "pr": "pending-merge",
-      "b24ui_sha": "pending-merge",
+      "pr": 432,
+      "b24ui_sha": "807d52efaf8a29cc23b97df3a047c082752bfad0",
       "decision": "port",
       "summary": "docs(showcase): add kleinezeitung.at (nuxt/ui) — PORT of the schema half; content half is a no-op for the same reason as cf5f15e3. Upstream's entry needed a wider viewport, a delay and an element removed before the screenshot was usable, so screenshotOptions gained width/cookies/removeElements and delay became optional. Ported because it is not decorative here: docs/modules/showcase.ts reads screenshotOptions and spreads it straight into capture-website, so the new keys reach the screenshotter as soon as the schema stops rejecting them, and the old schema REQUIRED delay so an entry wanting only width would have failed validation. DEVIATION recorded: while checking the widening would actually do something, found that delay was inert in this fork — our module hardcoded `delay: 2` AFTER the spread, overwriting any per-entry value, where upstream has no default there at all (which is why their schema could treat delay as the one real option). Porting the widening as-is would have shipped a schema advertising an option the module discards, the same shape of false claim as the MCP inputExamples that named non-existent examples. The fork's 2s default is worth keeping, so it moved BEFORE the spread: still a default, now overridable. One line, a deviation from the upstream commit rather than part of it. No showcase entry sets screenshotOptions today, so nothing changes behaviourally — the point is that schema and module now agree. Also refreshed .sync/dep-parity.json to this cursor; versions identical."
     }