Decouple tag identity from presentation (Tag.display_name)

[WilfordGrimley]

Summary

Decouples tag identity (Tag.name, immutable) from tag presentation
(Tag.display_name, freely editable) before more tag data accumulates on
top of the printing/tag consensus system.

1. Model: Tag.display_name — nullable CharField, additive migration
   0056_tag_display_name. name remains the machine key used by votes,
   tag_vote_statuses, Card.tags, and federation; the field's help text
   spells out "NEVER rename a Tag after creation."
2. Serialization: Tag.serialise() / SerialisedTag schema (Python +
   TS, quicktype-regenerated) gain displayName. Frontend renders
   displayName ?? name at every tag render site — queue strips
   (NoMatchReasonStrip, PrintingConfirmStrip, QueueTagQuestion),
   TagVotePicker chips, the tag filter UI, and the card detail modal —
   via a shared useTagDisplayName() hook backed by the already-cached
   useGetTagsQuery(). All API submissions/filters still send name.
3. Seeding: seed_no_match_reason_tags now also backfills
   display_name for its six tags ("Custom art", "Altered frame",
   "Upscaled", "AI art", "No collector line", "Non-English"), and
   seed_default_tags backfills it for the legacy "Full Art"/"Borderless"
   tags. Both are idempotent and only fill when display_name is
   currently None — a manually-edited display name is never clobbered on
   re-seed. No existing tag's name was renamed.
4. Indexing-pipeline check (no code changes): cardpicker/tags.py's
   Tags.get_tags() / match_tag_fuzzy() / extract() match filename
   bracket tokens against Tag.name and Tag.aliases only — never
   display_name — and Card.tags persists tag_object.name. So this
   stage changes nothing about indexing today. A future change to
   name's key style (not something this PR does) would still break two
   things this PR doesn't touch: stale .lower() filename-bracket matches
   against the old name, and already-persisted Card.tags snapshot arrays
   with no reconciliation migration path. Full writeup in
   docs/features/printing-tags.md's new Stage 5 section.
5. Docs: one-line additions to docs/federation-v1.md (tag name is
   the immutable interchange key; display_name never federates) and a new
   Stage 5 section in docs/features/printing-tags.md covering the model,
   serialization, hook, seeding, and the item-4 analysis in full.
6. Tests: serialization includes displayName (backend + a new
   TestGetTags case with a set display name), frontend fallback renders
   raw name when displayName is unset (new TagVotePicker.spec.ts
   case using a mock where only one tag has a display name), and seeding
   idempotence (both test_reason_tags.py and new test_default_tags.py
   assert re-running the seeder never overwrites a manually-set
   display_name).

Test plan

• Backend: pytest — 459 passed, 4 pre-existing unrelated failures
  (test_integrations.py::TestMTGIntegration::test_valid_url[moxfield],
  [moxfield_without_www], test_sources.py::TestUpdateDatabase::test_comprehensive_snapshot,
  test_upsert — same baseline as master, not touched by this change).
• Frontend: Jest 123/123 (excluding one pre-existing Node-version-gapped
  file, same as baseline), eslint 0 errors, prettier --check clean,
  npm run build succeeds (static export, all 8 pages).
• Playwright: 161/161, including the two new/updated tag-display-name
  cases.
• Manual visual check: screenshotted NoMatchReasonStrip,
  PrintingConfirmStrip, and TagVotePicker against mocks with real
  display names set — confirmed correct rendering (including the
  known-and-intended "Full art" → "Full Art" chip-label change) with no
  loading flicker or layout issues.

Holding as draft for review per task instructions — not merging.

🤖 Generated with Claude Code

https://claude.ai/code/session_016i9S7LQsCL3FGaih3ZTRBJ