test: add production-scale reorder fixture and deterministic failure reproduction

[devin-ai-integration]

Summary

Closes #976 (first sub-issue of epic #975).

Adds the deterministic, production-scale reorder fixture and the regression scaffolding the rest of the epic builds on:

• backend/app/reorder_fixture.py — build_fixture_spec() / seed_reorder_fixture() / purge_reorder_fixture() plus a CLI (python -m app.reorder_fixture [--purge]). Deterministic IDs in reserved high ranges (9_100_000+ categories, 9_200_000+ images) and RF- prefixed names; 80 sibling categories, 600 sibling images, mixed root scope, 4-level nested scope, and duplicate sort_order values in every scope. Seeding purges first, so it is idempotent with no manual cleanup.
• backend/tests/test_reorder_fixture.py — spec tests (always run) + PostgreSQL integration tests gated on REORDER_FIXTURE_DATABASE_URL: idempotent reseed, purge, and a full 80-category authoritative-order round-trip through the real reorder_categories endpoint. Two xfail(strict=True) regressions document current bugs and will flip to failures when fixed:
  • mixed category/image reorder is not atomic (one half commits when the other fails → Implement an atomic, revisioned tile-order API with set-based updates #978);
  • a stale second-tab submission silently last-write-wins instead of conflicting (→ Implement an atomic, revisioned tile-order API with set-based updates #978).
• .github/workflows/ci.yaml — backend-tests now runs a postgres:16-alpine service, applies Alembic migrations, and sets REORDER_FIXTURE_DATABASE_URL so the integration tests run in CI.
• frontend/tests/helpers/reorderFixture.ts — mirrored deterministic generators + createDeferred() for injecting latency into category/image persistence and refreshes independently.
• frontend/tests/components/SortableTileGridReorderRegression.test.tsx — passing baselines (695-tile render, 20 consecutive awaited reorders) and three it.fails(...) regressions asserting the desired behaviour:
  • a second drop during an in-flight save is silently discarded (reorderInFlightRef early return);
  • a stale refresh response overwrites a newer saved order;
  • partial category/image persistence.
• docs/reorder-fixture.md — fixture-loading workflow, scenario mapping, and observed timings at fixture scale (linked from README).

Needs human verification

None — but note the navigation-away-during-save and reload-and-compare journeys are documented as browser-level (Playwright) scenarios to be driven against a fixture-seeded environment; they are not automated in this PR.

Testing

• backend: poetry run pytest — 940 passed, 2 xfailed, coverage 86%.
• frontend: npm test — 1323 passed, 3 expected fail; npm run format:check clean.

Link to Devin session: https://app.devin.ai/sessions/414754bd88374dc693f47100ac269e78

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment, CI, and merge conflict monitoring

[devin-ai-integration]

Devin Review found 3 potential issues.

[devin-ai-integration]

Devin Review found 2 new potential issues.

[devin-ai-integration]

Devin Review found 3 new potential issues.

[devin-ai-integration]

Devin Review found 2 new potential issues.