perf: implement Redis caching layer for gist queries

[anonfedora]

Title: perf: implement Redis caching layer for gist queries

Description:
This PR implements a Redis caching layer for gist queries to reduce read load on PostgreSQL and improve response times for repeated queries.

Changes

Core Implementation

• Added ioredis dependency for Redis client connection
• Created CacheModule and CacheService with graceful degradation when Redis is unavailable
• Cache key format:
  • Nearby queries: gist:nearby:{lat}:{lon}:{radius}:{limit}
  • Single gist: gist:one:{id}
• TTL configuration:
  • Nearby queries: 60 seconds
  • Single gist: 300 seconds (5 minutes)

Service Integration

• Modified GistsService.findNearby() to cache results (excludes paginated queries with cursor)
• Modified GistsService.findOne() to cache individual gists
• Added invalidateNearbyCache() method called on gist creation
• Cache invalidation uses pattern-based deletion for geographic area

Configuration

• Added REDIS_URL environment variable to .env.example
• Integrated CacheModule into AppModule and GistsModule
• Redis connection includes retry strategy and error handling

Testing

• Unit tests: 9 tests for CacheService covering initialization, get/set/delete operations, metrics, and graceful degradation
• Integration tests: Added to gists.e2e.spec.ts to verify graceful degradation when Redis is unavailable
• All existing unit tests pass (67/67)

Acceptance Criteria Met

✅ First query misses cache, hits DB, populates cache
✅ Second identical query returns from cache (no DB hit)
✅ New gist creation invalidates nearby cache for that area
✅ Redis unavailability does not crash the application
✅ Cache hit rate is logged/metrics available
✅ All existing tests pass

Technical Details

Graceful Degradation

• If REDIS_URL is not configured, caching is disabled and all queries hit the database
• Redis connection errors are caught and logged; service continues without caching
• No application crashes due to Redis unavailability

Cache Invalidation Strategy

• When a new gist is created, all nearby cache keys for that geographic area are invalidated
• Uses pattern matching: gist:nearby:{lat:.4f}:{lon:.4f}:*
• Ensures stale data is not served after new gists are added

Metrics

• CacheService tracks hits, misses, and hit rate percentage
• Metrics can be retrieved via getMetrics() method
• Useful for monitoring cache effectiveness

Testing Strategy

Unit Tests

• Mock Redis client to test service logic without actual Redis
• Test graceful degradation scenarios
• Verify metrics tracking

Integration Tests

• Test that application works correctly when Redis is unavailable
• Verify no crashes or errors during cache operations

CI/CD

• All tests pass in CI environment
• Linting passes with no errors
• Build succeeds with TypeScript compilation

---

Closes #25

[snowrugar-beep]

Nice work @anonfedora! 🎉 CI is green across the board and the Redis caching layer is a really solid improvement — clear TTLs, graceful fallback when REDIS_URL is unset, and hit-rate metrics baked in. Thanks for tackling #25, merging this in now.