feat(e2e): add Playwright browser test suite for web routes

solitaire_server/e2e/:
- smoke.spec.js: verifies /play-classic loads, exposes window.__FERROUS_DEBUG__
  bridge, keyboard parity (Space=draw, U=undo), debug failure report, and
  replay payload builder exports schema-v2 moves.
- gameplay_review.spec.js: HUD/controls render check, stock-click + undo
  player flow, draw-mode toggle, autonomous play invariant batch, and
  cycle-detection regression guard.
- cycle_metrics.js: headless cycle-rate analysis tool; run via
  `npm run review:cycles` with configurable policy, game count, and
  thresholds. Regression gate baked into package.json scripts.
- playwright.config.js: targets the local server at http://localhost:8080.
- package.json / package-lock.json: @playwright/test 1.60.0.

.gitea/workflows/web-e2e.yml:
- Runs on pushes to solitaire_server/, solitaire_wasm/, solitaire_core/,
  or Cargo changes. Starts the server binary, waits for /health, runs
  the full Playwright suite, uploads test-results/ on failure.

docs/testing-architecture.md: documents the three-tier test strategy
  (unit → Playwright smoke → cycle regression) and the __FERROUS_DEBUG__
  bridge contract.

scripts/update_quaternions_deps.sh: helper to bump the Quaternions
  registry deps (klondike, card_game) by version and run the full
  safety gate including deterministic replay checks.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

docs/testing-architecture.md

@@ -0,0 +1,115 @@
+# Testing Architecture — Engine-first Validation
+
+Ferrous Solitaire validation is split into three layers with clear ownership:
+
+1. **Rust unit tests (`solitaire_core`)**
+   - move generation and legality
+   - deal generation determinism
+   - scoring and penalties
+   - undo semantics
+   - win detection
+
+2. **Engine integration tests (`solitaire_wasm` debug API)**
+   - autonomous game execution without UI/pointer simulation
+   - invariant checks after every move
+   - deterministic seed replay
+   - high-volume seeded runs (including long-running soak tests)
+
+3. **Playwright UI tests**
+   - verify rendering vs engine state
+   - drag/drop and keyboard UX behavior
+   - responsive layout behavior
+   - browser-compatibility checks
+
+## Source of truth
+
+The Rust engine is authoritative. Browser tests must interact with the game via
+debug API hooks, not via pixel/OCR solving or hardcoded screen coordinates.
+
+## Debug API surfaces
+
+Two automation surfaces are exposed:
+
+- `solitaire_wasm::SolitaireGame` methods:
+  - `debug_snapshot()`
+  - `debug_legal_moves()`
+  - `debug_move_history()`
+  - `debug_apply_legal_move(index)`
+  - `debug_apply_move_json(json)`
+- Browser bridge on `game.html`:
+  - `window.__FERROUS_DEBUG__.snapshot()`
+  - `window.__FERROUS_DEBUG__.legalMoves()`
+  - `window.__FERROUS_DEBUG__.moveHistory()`
+  - `window.__FERROUS_DEBUG__.applyLegalMove(index)`
+  - `window.__FERROUS_DEBUG__.applyMove(move)`
+  - `window.__FERROUS_DEBUG__.failureReport()`
+  - `window.__FERROUS_DEBUG__.runAutoplay(options)`
+
+## Required failure payload
+
+Every automation failure should capture:
+
+- seed
+- move history
+- current game state
+- screenshot
+- browser trace
+- console logs
+
+`failureReport()` provides the engine-side fields (`seed`, `moveHistory`,
+`currentState`) so UI harnesses only need to attach browser artifacts.
+
+## Execution guidance
+
+- Fast verification:
+  - `cargo test -p solitaire_core -p solitaire_wasm`
+- Full verification:
+  - `cargo test --workspace`
+  - `cargo clippy --workspace -- -D warnings`
+- Long unattended soak:
+  - `cargo test -p solitaire_wasm debug_api_autonomous_thousands_seed_soak -- --ignored`
+
+### Browser e2e harness
+
+The Playwright suite lives under `solitaire_server/e2e/` and boots
+`solitaire_server` via Playwright `webServer` config.
+
+- Install + run:
+  - `cd solitaire_server/e2e`
+  - `npm ci`
+  - `npx playwright install chromium`
+  - `npm test`
+- Cycle metrics batch run:
+  - `cd solitaire_server/e2e`
+  - `npm run review:cycles -- --games 1000 --steps 350 --policy baseline --max-visits 1 --out /tmp/cycle-baseline.json`
+  - `npm run review:cycles -- --games 1000 --steps 350 --policy loop_aware --max-visits 2 --out /tmp/cycle-loop-aware.json`
+  - `npm run review:cycles:regression` (thresholded gate, writes `test-results/cycle-regression.json`)
+  - `npm run review:cycles:candidate` (loop-aware candidate run, writes `test-results/cycle-candidate.json`)
+
+### Cycle-risk regression baseline and guardrails
+
+- Current regression gate command:
+  - `npm run review:cycles:regression`
+  - config: `games=240`, `steps=350`, `policy=baseline`, `max-visits=1`
+- Current guardrail thresholds:
+  - `all.cycle_rate_pct <= 86`
+  - `draw1.cycle_rate_pct <= 76`
+  - `draw3.cycle_rate_pct <= 95`
+  - `all.win_rate_pct >= 14`
+  - zero invariant/apply/page/console issue counts
+- Baseline sample (240 games):
+  - overall: `win_rate=15.8%`, `cycle_rate=84.2%`
+  - draw-one: `win_rate=25.8%`, `cycle_rate=74.2%`
+  - draw-three: `win_rate=5.8%`, `cycle_rate=94.2%`
+- Candidate loop-aware sample (240 games, lookahead via simulated move + restore):
+  - overall: `win_rate=20.4%`, `cycle_rate=32.5%`
+  - draw-one: `win_rate=33.3%`, `cycle_rate=16.7%`
+  - draw-three: `win_rate=7.5%`, `cycle_rate=48.3%`
+  - no invariant/apply/page/console issues in the sampled run
+- Additional 500-game candidate soak:
+  - overall: `win_rate=20.2%`, `cycle_rate=28.6%`, `step_budget=51.2%`
+  - draw-three remains the dominant risk (`cycle_rate=45.2%`)
+- Fix applied: cycle metrics regression now supports explicit
+  `max_step_budget_rate_*` thresholds. Candidate command now enforces
+  `max_step_budget_rate_all <= 60` to prevent silent drift from cycles into
+  step-budget stalls.