Port ChangeQueryModal/Toasts/a11y/InvalidIdentifiers/SelectVersionSection clusters onto DisplayPage

[WilfordGrimley]

Description

Wave 2 of issue #272's parity work (following wave 1's PR #395): un-skips and
ports 44 Playwright tests from the classic /editor grid's coverage onto the
unified DisplayPage at /editor (PR #389's own swap), stacked on
proposal-h-editor-display-route-swap (PR #389) the same way wave 1 was.

Ported clusters (44 tests)

1. SelectVersionSection.spec.ts (7 tests) - this file was never actually
   classic-editor-only: it already exercised the unified page's own rail (via
   openSelectVersionSection, added to test-utils.ts back when
   DisplayLeftRailFidelity.spec.ts landed) since issue Select Version section build (blocked on PR #116) #167 shipped, well
   before the route swap. It picked up the swap's blanket per-file skip
   marker anyway and was deliberately left for this wave. Un-skipped with no
   other change - all 7 pass as-is.
2. ChangeQueryModal.spec.ts (9 of 12 tests) - ChangeQueryModal.tsx is
   unchanged and still globally mounted (Modals.tsx); reached via the
   shared CardSlotContextMenu's "Change Query" action (right-click a sheet
   slot), the same context menu DisplayPage.tsx already wires up
   ("no new action, just three more ways to reach it on /display" per that
   file's own comment). New openDisplayChangeQueryModal helper in
   test-utils.ts.
3. Toasts.spec.ts (3 of 9 tests, the rest were already passing) - the
   DFCPairs/importSites/sampleCards error-toast tests used
   openImportTextModal/getAddCardsMenu only as a means to mount
   ImportText.tsx/ImportURL.tsx, whose own RTK Query hooks fire on
   mount, not on any click. DisplayPage's empty landing mounts both
   unconditionally (ImportColumns, URL accordion defaultActiveKey="url")
   • all three fetches already fire on plain page load, same as the
     already-passing /2/cards//2/sources tests beside them.
4. GeneralUIAccessibility.spec.ts (3 of 4 tests) - the per-slot "more
   options" a11y checks retarget onto PagePreview.tsx's own per-slot menu
   cue (page-preview-slot-menu-cue, "Open card menu" - the sheet's closest
   equivalent, reserving the same corner classic CardSlot's 3-dot button
   occupied). The editor-page console-warnings test ports directly.
5. InvalidIdentifiersStatus.spec.ts (5 tests) + its visual companion (1
   test) - InvalidIdentifiersStatus.tsx/InvalidIdentifiersModal.tsx are
   both unchanged; DisplayPage.tsx's own D13 comment confirms the
   landing/search-bar mount (which these tests exercise) already works -
   only the separate right-rail Status row (issue /display parity inventory: verified remaining gaps vs /editor (switchover checklist) #272 item 2's own
   remaining scope) is missing.
6. SearchSettings.visual.spec.ts (1 test) - SearchSettings.tsx is
   unchanged, relocated into the right rail unmodified. New
   openDisplaySearchSettingsModal helper (opens the gear button/right
   rail first, when it isn't already inline).

Deviations from a clean port

• 3 of ChangeQueryModal.spec.ts's 12 tests dropped: the multi-slot
  DFC tests ("plural text when multiple front slots are selected",
  "updates all fronts and backs for a multi-slot selection", "not shown
  when any one of the selected slots already has the DFC back query")
  depended on the classic grid's checkbox multi-select +
  SelectedImagesRibbon's own "Change Query" trigger - bulk multi-select
  has no equivalent on the unified page (see "Parked clusters" below). The
  9 single-slot DFC prompt/submission/condition/fuzzy-search/checkbox-reset
  tests are unaffected and fully prove the same underlying logic.
• GeneralUIAccessibility.spec.ts's "comfortable touch-target size"
  (>=40x40px) assertion dropped, not weakened: PagePreview.tsx renders
  the sheet near print-scale (millimetre-driven sizing), and the menu cue
  measured ~11x11px there - a real, structural difference in what the
  element IS (a small on-page-preview affordance vs. a full-size toolbar
  button), not an accessibility regression. The focus-visible-style half of
  that same test is unaffected and still verified.
• GeneralUIAccessibility.spec.ts's two "Editor - mobile layout" tests
  dropped (left-panel/right-panel 50/50-split-vs-stacked): they verified
  the same "mobile scroll affordances -> Mobile /display: fit-to-width sheet + bottom-sheet drawer for card actions (#231 blocker) #266's responsive layer"
  replacement issue /display parity inventory: verified remaining gaps vs /editor (switchover checklist) #272's own body already lists as intentional, not a
  gap. The unified page's rails are off-canvas drawers below their own
  breakpoints regardless of viewport - there's no persistent 50/50 split to
  ever "stack" in the first place.
• CardImageStates.spec.ts investigated, left skipped with an improved
  reason - not a route-swap casualty after all: its
  card-image-error-placeholder/card-image-slow-load-hint testids only
  exist in Card.tsx. PagePreview.tsx's sheet slots are their own
  component with their own loading/failed states (already covered by
  DisplaySlotStates.spec.ts) but a plain, unwrapped <img> for the
  "candidate resolved, but its own image fetch 404s or hangs" case - no
  error-placeholder swap, no slow-load hint, at all. This is a genuine,
  confirmed gap (new UI work, not test-porting), not something this wave
  can port.
• DeckbuilderConfirmAffordance.spec.ts investigated, left skipped -
  its rail-header mount was intentionally REMOVED and superseded by
  ConfidenceElement.tsx (DisplayPage.tsx's own D14 comment). Its only
  remaining reachable mount (the Select Version confirm ribbon) is already
  covered by SelectVersionSection.spec.ts's own "carries the confirm
  ribbon" test, ported above - porting this file too would just duplicate
  that coverage against a UI surface (the auto-shown top-level badge) that
  no longer exists on this page.

Parked clusters (skip left untouched, not ported or dropped)

I could not find an authoritative "cluster N" list in issue #272 itself
(its own checklist is 8 numbered GAPS, not test clusters) - the mapping
below is my own inference from the issue body's "Intentional replacements,
NOT gaps" list plus source investigation. Flagging this so it can be
corrected if it doesn't match the intended list:

• visual/CommonCardback.visual.spec.ts - issue /display parity inventory: verified remaining gaps vs /editor (switchover checklist) #272's own body:
  "CommonCardback mount -> CardbackToolbarButton" is an intentional
  replacement, not a gap.
• SelectedImagesRibbon.spec.ts (bulk multi-select) - issue /display parity inventory: verified remaining gaps vs /editor (switchover checklist) #272 item
  6, still not built; no DisplayPage surface exists to point this at.
• ProjectEditorMobileScroll.spec.ts - issue /display parity inventory: verified remaining gaps vs /editor (switchover checklist) #272's own body: "mobile
  scroll affordances -> Mobile /display: fit-to-width sheet + bottom-sheet drawer for card actions (#231 blocker) #266's responsive layer" is an intentional
  replacement.
• PDFGenerator.spec.ts + PagePreview.spec.ts +
  PostExportContributionPrompt.spec.ts - the PDF-generation surface
  split across two entry points (the classic embedded PDF tab, now gone -
  DisplayPage's own center sheet region is the preview now per the owner's
  addendum - vs. the separate /print route, generation/settings only).
  Where this coverage should live now is a placement decision, not a
  mechanical port.

Deferred to a future wave (not owner-parked - time/complexity only)

• GridSelectorModal.spec.ts (23 tests) + GridSelectorModalVariants.spec.ts
  (7 tests) - the modal is reachable on DisplayPage (e.g. via
  CardbackToolbarButton), but mapping this cluster's grouping/filter/
  title-count assertions onto a genuinely reachable instance is real,
  non-mechanical work I didn't get to this wave.
• CardSlot.spec.ts (25 tests) + visual/CardSlot.visual.spec.ts -
  large per-slot interaction cluster (image cycling arrows, per-slot
  context menu, badges) that needs redesigning around the sheet+rail
  architecture rather than a 1:1 retarget.

Saying this plainly rather than silently shrinking wave 2's scope: these
two clusters are real, non-trivial remaining work, not confirmed gaps or
owner decisions.

Verification

• All 44 newly-unskipped tests green, run individually per-cluster and
  together.
• Wave 1's marquee regression files re-run: ImportXML.spec.ts (16),
  chunkErrorRecovery.spec.ts, DisplayLeftRailFidelity.spec.ts - all
  pass.
• Full suite (npx playwright test, excl. perf): 325 total, 225 passed, 96
  skipped (unchanged - only intra-file skips converted to real tests, no
  file went from unskipped to skipped), 4 failed on the first full run
  (Navbar.spec.ts's Sign-out test, UnsavedWorkGuard.spec.ts's
  editor->/print test, visual/CardDetailedViewModal.visual.spec.ts,
  visual/DynamicLogo.visual.spec.ts) - all 4 re-run individually and
  passed; none touch a file this PR changed. DynamicLogo.visual.spec.ts's
  failure in the full run is this session's own dedicated-port workaround
  (its MSW handler hardcodes localhost:3000), documented in wave 1's own
  PR as the same known artifact.
• npx tsc --noEmit clean.
• npx jest --silent - 568/568 passing (unchanged; no src/ files
  touched).
• npx prettier@2.7.1 --check clean on all changed files.

Environment note: local runs used a temporary, uncommitted
playwright.config.ts port override (3411) per docs/lessons.md's
worktree-port-collision entry, reverted before every commit.

Open items for the owner

1. The "cluster N" mapping in the Parked section above is inferred, not
   sourced from an explicit list - please correct if it's wrong.
2. GridSelectorModal.spec.ts/GridSelectorModalVariants.spec.ts/
   CardSlot.spec.ts (55 tests total) remain deferred, not parked - real
   scope for a wave 3.
3. CardImageStates.spec.ts's gap (no image-fetch-error/slow-load UX on
   the sheet) may be worth its own tracked issue, separate from /display parity inventory: verified remaining gaps vs /editor (switchover checklist) #272.

Checklist

• I have installed pre-commit and installed the hooks with
  pre-commit install before creating any commits.
• I have updated any related tests for code I modified or added new
  tests where appropriate. (This PR is the test update.)
• I have manually tested my changes as follows: see Verification
  above.
• I have updated any relevant documentation or created new
  documentation where appropriate. (No docs/ file tracks per-wave
  test-porting status, matching wave 1's own precedent.)

---

Stacked on proposal-h-editor-display-route-swap (PR #389, held open,
parity-blocked), same stacking as wave 1's PR #395.