Session 44: Make it visible — VYNDR archetype names, grade intel, schedule fix, landing page (2061 tests)

Frontend + wiring only. Wires existing backend into the pages users see.

- VYNDR Original archetype rename (41) across archetypeService.js + lib/
  archetypes.js + ArchetypeBadge, each keeping legacyName (resolves stale data).
  Judge -> BOMBER. Old POWER PULL slot -> WHIFF strikeout-artist pitcher.
- BACKEND_HANDOFF.md: canonical frontend<->backend data contract.
- Grade card intel: scan/page.tsx now forwards the engine's intel fields
  (season_avg/form/usage/matchup_grade/archetype/...) into mapScanToGradeResult
  -> STAT CONTEXT + VYNDR INTELLIGENCE sections populate. The chain already
  preserved them (tierGating + /api/scan spread); the page was dropping them.
- Schedule freshness: slateAdapter.isRelevantGame drops completed games >24h
  old; Slate.filteredGames applies it. (TTL already 60s.)
- Landing: Features.tsx rewritten to user-facing copy (no Point-biserial/Zone
  14/ABS/Phi-coefficient).
- Depth chart Next proxies added (/api/stats/lineup|depth|cascade) - were 404.
- GameCard swap DEFERRED (Kev): legacy on-demand card stays as a bridge until
  the snapshot pipeline populates the grades cache; vyndr/GameCard swaps in then.

Backend 2045 -> 2061 tests (+16), 167 suites. Web build clean (exit 0).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

web/src/components/Features.tsx

@@ -1,33 +1,33 @@
 const FEATURES = [
   {
     icon: '◆',
-    title: 'Multi-dimensional player archetypes',
-    body: 'Players aren\'t one thing. Our model scores every dimension — pitcher discipline, batter approach, NBA usage shape — and blends them per matchup.',
+    title: 'Player DNA archetypes',
+    body: 'Every player has a prop fingerprint. We classify it, show you which props are reliable vs volatile, and grade accordingly.',
   },
   {
     icon: '↻',
-    title: 'Auto-calibrating engine',
-    body: 'Every resolved grade trains the next one. Point-biserial weight tuning, per-stat calibration, blind-spot detection. The model improves itself.',
+    title: 'Self-improving model',
+    body: 'Every resolved grade makes the next one sharper. The engine learns what works and corrects what doesn\'t.',
   },
   {
     icon: '⚡',
-    title: 'Beat reporter intelligence',
-    body: 'Lineup intel from the people closest to the team — 30 minutes before tip. Trust-tiered, redistribution-aware, line-correlated.',
+    title: 'Lineup intel before tip-off',
+    body: 'Real-time lineup and injury data from trusted sources, giving you the edge before the books adjust.',
   },
   {
     icon: '⊘',
     title: 'Kill conditions',
-    body: 'We don\'t just grade the prop. We tell you what kills it. Six hard checks per read: minutes, sample, fatigue, blowout risk, splits, line conflict.',
+    body: 'Six hard checks on every prop. If minutes, fatigue, blowout risk, or splits say no, we flag it — even on an A grade.',
   },
   {
     icon: '∿',
     title: 'Parlay correlation math',
-    body: 'Phi-coefficient analysis catches the legs that secretly fight each other. The books love correlated unders. We surface them.',
+    body: 'We catch the legs that secretly fight each other. The books love correlated unders — we surface them before you tap.',
   },
   {
     icon: '⌧',
-    title: 'ABS intelligence (MLB)',
-    body: 'The automated strike zone changes everything. Per-pitcher, per-batter discipline scoring. Zone 14 framing loss. Challenge math.',
+    title: 'Deep pitcher-batter matchups',
+    body: 'We score discipline, contact quality, and zone tendencies for every MLB matchup. Not just ERA.',
   },
   {
     icon: '◯',