feat(cache): implement Redis caching layer for hot-read endpoints

Add cache-aside pattern for listing detail, search results, market
analytics (4 endpoints), and user profile queries. Cache invalidation
on all write mutations. Prometheus cache_hit_total/cache_miss_total
metrics with resource labels.

- CacheService: getOrSet, invalidate, invalidateByPrefix (SCAN-based)
- TTLs: listing 5m, search 1m, market 30m, profile 10m
- All 230 tests passing (13 new cache tests + 6 updated handler tests)

Co-Authored-By: Paperclip <noreply@paperclip.ing>

apps/api/src/modules/metrics/metrics.module.ts

@@ -47,6 +47,18 @@ import {
       buckets: [0.005, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1],
     }),
 
+    // ── Cache Metrics ──
+    makeCounterProvider({
+      name: 'cache_hit_total',
+      help: 'Total number of cache hits',
+      labelNames: ['resource'],
+    }),
+    makeCounterProvider({
+      name: 'cache_miss_total',
+      help: 'Total number of cache misses',
+      labelNames: ['resource'],
+    }),
+
     // ── Business Metrics ──
     makeCounterProvider({
       name: 'listings_created_total',