Lighthouse regression: 90 → 74-82

[thefiredev-cloud]

Entry 998KB + common 870KB. 226KB unused JS. Medium priority.

[thefiredev-cloud]

🔧 CTO Analysis & Fix Plan

Working on branch: \ix-lighthouse-40\

Root Cause Analysis

After reviewing the codebase, here's what's causing the regression:

Issue	Cause	Impact
Entry bundle 998KB	SummaryCard eagerly imported via MessageBubble	~150-200KB in critical path
226KB unused JS	AsyncStorage (@react-native-async-storage) + expo-haptics bundled for web	~226KB in entry/common
tRPC queries blocking paint	\coverageByState\ query fires immediately on mount	Delays FCP by network RTT
Common bundle 870KB	No web shims for AsyncStorage/Haptics	Both package full native deps

Fix Plan

1. Lazy-load SummaryCard - Wrap in \React.lazy()\ inside \MessageBubble.tsx\ so the card + its deps (reanimated, DrugInfoCard) don't block initial render
2. Add AsyncStorage web shim - Create \lib/async-storage-web-shim.ts\ that uses \localStorage\ + register in \metro.config.js\ (~200KB saved)
3. Add expo-haptics web shim - Create \lib/haptics-web-shim.ts\ pointing to our existing \lib/haptics.ts\ web logic (~26KB saved from expo-haptics native code)
4. Defer tRPC queries - Add \enabled: false\ +
   etworkMode: 'always'\ with \setTimeout\ defer pattern in \use-filter-state.ts\ so \coverageByState\ fires after first paint

Expected Outcome

• Bundle: entry ~800KB, common ~650KB
• FCP: 3.3s → ~2.0s
• Lighthouse: 74-82 → 90+

Starting implementation now.