Proposal H pane migration: build the display rail's remaining sections

[WilfordGrimley]

Closes #164

Summary

Builds the four rail sections DisplayPage.tsx (/display) still had as
labeled stubs, per docs/proposals/proposal-h-unified-display-page.md's
§5 component-mapping table (the "left-panel unification" step of the pane
migration):

• Attributes — the tri-state attribute-chip taxonomy
  (attributeChips.ts), now voted on via a shared useTagVoting hook
  extracted from AttributeChipPanel.tsx (unchanged public behavior,
  AttributeChipPanel.test.tsx still green). The rail renders the same
  Chip/vote-submission logic in a plain vertical stack instead of a ring
  around a card image, fetching this slot's tag consensus via the existing
  APIGetTagConsensus endpoint (same one bleedPriorResolution.ts already
  uses for the export-time bleed prior).
• Print Options — the per-card bleed override (Auto / Force bleed /
  Force trimmed), reusing the exact projectSlice selector/action and a
  newly-exported isBleedNormalizationEligible (from PDF.tsx) so the
  classic PDF tab's BleedOverrideSettings panel and this rail section
  share one eligibility rule instead of two that could drift. (The
  original stub's "blocked on Proposal B PR-2" note is stale — that PR
  shipped; see docs/features/pdf-generator.md's Proposal B status.)
• Artist — inherits ArtistSupportLink directly, exactly as
  docs/features/artist-support-links.md anticipated this surface would
  once built.
• Slot Actions — the same getCardSlotMenuActions list CardSlot.tsx's
  3-dot dropdown/context menu use, rendered as a plain action list rather
  than a dropdown overlay, per the design doc's own instruction for this
  section.

Scope

This is #164 only (pane migration + left-panel unification). Explicitly
NOT touched, per the task's own scope fence:

• Proposal G save integration #165 (Proposal G save integration) — untouched.
• Post-export contribution prompt #166 (post-export contribution prompt) — untouched.
• Select Version section build (blocked on PR #116) #167 (Select Version section build) — separately HELD on PR Proposal H: Select Version section unified spec (HOLD, spec only) #116/issue
  PR #116 blocked on formatting/type-check failure #140; the existing Choose Image section (candidate/version picker) was
  already built in an earlier PR and is unchanged here.
• §6 step 3 (switchover — making /display the default nav entry over
  /editor) and step 4 (retiring /editor) — not part of "left-panel
  unification" by name; flagged as an open decision below rather than
  guessed at.
• The tablet off-canvas drawer / mobile bottom-sheet interaction patterns
  (§3) — still the plain-document-flow fallback below md, unchanged.

Performance benchmark (owner condition: before AND after)

Reused the existing scroll/virtualization benchmark harness landed by PR
#178 (frontend/tests/perf/display-scroll.bench.spec.ts,
playwright.perf.config.ts) unmodified — a 120-card/15-sheet deck,
scrolled through in steps under 4x CPU throttling, sampling fps/heap/
mounted-<img>-count/long-tasks. Ran it on the untouched baseline first
(before any edit in this branch), then again after landing this change:

Metric	Before (baseline)	After (this PR)
Avg fps (target ~60)	58.7	59.8
p95 frame time	23.0ms	22.7ms
Peak JS heap	256.5 MB	256.5 MB
Max simultaneously-mounted <img> tags (of 120)	16	16
Long tasks (>50ms) during scroll	1 (86.0ms)	0

No regression — numbers are effectively flat, within normal run-to-run
noise. This is the expected result, not just a lucky one: this PR's
changes are confined to the rail (mounted per-slot on selection), and the
benchmark's own scroll flow never selects a slot — it exercises the sheet
stack's virtualization, which this PR doesn't touch. A flat result
confirms the rail work didn't regress sheet scroll performance; it isn't
evidence about the rail's own render cost under this harness, which wasn't
designed to measure that.

Verification

• npx tsc --noEmit — clean.
• npx jest — 43 suites / 402 tests passed (includes
  AttributeChipPanel.test.tsx, unchanged behavior after the
  useTagVoting extraction).
• npx playwright test (full suite, 273 tests) — 272 passed, 1
  (ImportCSV.spec.ts) failed under full-suite resource contention and
  passed cleanly in isolation (pre-existing flake, unrelated to this
  change — not touched by this PR).
• npx prettier@2.7.1 --check on every changed frontend file — clean.
• Updated tests/DisplayPage.spec.ts: fixed 3 existing assertions that
  referenced the old Attributes stub text, added tagConsensusTwoUnresolvedTags
  to every slot-selecting test's mocks (the Attributes section fetches on
  slot-select regardless of whether it's expanded — same eager-fetch
  precedent ChooseImageSection already set), and added 4 new tests (one
  per newly-built section).

Docs

• docs/proposals/proposal-h-unified-display-page.md — status line
  updated (was still marked "zero feature code," despite Proposal H, Step 1: /display route shell behind a feature flag #87/Proposal H, Step 2 PR 2a: real candidate/version picker in the display rail #96/Proposal H, Step 2 PR 2b: requested-printing badge + Confirm affordance #102/
  Proposal H, item 1: missing-image slots keep their card name #104/Proposal H, item 2: inline PDF export runs the real pipeline on /display #109/Item (c): requested-printing badge on editor slots #110/Proposal H, item 3: flat vertical scroll + sheet-level virtualization #115 having shipped since; now lists what's built vs.
  still open).
• docs/README.md — Proposal H's status row: HOLD → PARTIAL.
• docs/features/artist-support-links.md — the rail is now a third real
  surface, not an anticipated follow-on.
• docs/features/pdf-generator.md — notes the shared
  isBleedNormalizationEligible export.
• docs/features/printing-tags.md — corrected a stale reference to
  Chip's location after the attributeChipRender.tsx extraction.

Open items for the owner

1. Switchover (§6 step 3: making /display the default nav entry point
   over /editor) isn't part of "left-panel unification" by name — left
   undone and unguessed. Separate call needed on when/whether to flip it.
2. No dedicated ineligible-source (e.g. non-Google-Drive/local-file)
   Playwright coverage for the Print Options section — every existing
   card fixture used in this PR's tests is Google-Drive-sourced, so the
   ineligible branch is exercised only by the pre-existing
   isBleedNormalizationEligible/BleedOverrideSettings coverage, not a
   new dedicated test. Flagging so it's a known, not silent, gap.

[WilfordGrimley]

Wiki-maintenance check (CLAUDE.md's Tooling rules): confirmed against .github/wiki-publish-map.json and docs/documentation-process.md rather than assumed. docs/proposals/ is explicitly excluded from every publish target, so the proposal doc edit needs no wiki action. docs/features/artist-support-links.md and docs/README.md aren't in the publish map either. docs/features/pdf-generator.md and docs/features/printing-tags.md ARE mapped and will auto-publish to the wiki via the existing docs-wiki-publish.yml CI on merge to master — no manual wiki edit needed (docs/documentation-process.md: "Never hand-edit a wiki page this system manages").

Merge-time checklist:

• No manual wiki edit required for this PR — see note above.

Full report (six-field standing format): https://github.com/ProxyPrints/ProxyPrints.github.io/blob/report-relay-proposal-h-pane-migration-99146/docs/reports/2026-07-19-proposal-h-pane-migration.md