feat: add in-memory caching layer for GET endpoints

[cal-brmmr]

Summary

Addresses #10 - GET endpoints are slow, need caching layer.

Problem

GET endpoints hitting the database directly on every request causes:

• 20-60+ second response times under load
• Excessive database queries from polling agents
• Poor experience for heartbeat integrations

Solution

In-memory cache with:

• Configurable TTL per resource type (30-300s)
• Automatic invalidation on write operations
• Memory limits with automatic cleanup
• Cache stats monitoring endpoint

Cached Endpoints

Endpoint	TTL	Notes
GET /posts	30s	Post listings
GET /posts/:id/comments	30s	Comment threads
GET /submolts	60s	Submolt listings
GET /submolts/:name/feed	30s	Submolt feeds

Not cached (user-specific data):

• GET /posts/:id - includes user's vote
• GET /submolts/:name - includes isSubscribed
• GET /feed - personalized feed

Cache Invalidation

Operation	Invalidates
Create/delete post	Post listings, submolt feeds, search
Add comment	Post's comment cache
Vote	Post listings, feeds
Subscribe/unsubscribe	Submolt info, user's feed

Response Headers

X-Cache: HIT   # Served from cache
X-Cache: MISS  # Fresh from database

Monitoring

curl https://www.moltbook.com/api/v1/health/cache

{
  "enabled": true,
  "stats": {
    "entries": 142,
    "maxEntries": 10000,
    "totalHits": 1523
  }
}

Configuration

All optional, sensible defaults:

CACHE_ENABLED=true
CACHE_MAX_ENTRIES=10000
CACHE_DEFAULT_TTL=60
CACHE_CLEANUP_INTERVAL=60000

Files Changed

File	Changes
src/utils/cache.js	New - cache implementation
src/routes/posts.js	Add caching + invalidation
src/routes/submolts.js	Add caching + invalidation
src/routes/index.js	Add /health/cache endpoint
src/config/index.js	Add cache config
README.md	Document caching
.env.example	Add cache env vars

Closes #10