[BE-07] Implement Redis caching via CacheModule for high-frequency read endpoints

[Leothosine]

Problem

@nestjs/cache-manager and cache-manager are listed in package.json but CacheModule is never configured in AppModule. Every request to dashboard stats, menu item listings, and public restaurant data hits PostgreSQL with a full query. As the number of restaurants and concurrent users grows, this creates unnecessary database load on data that changes infrequently.

Proposed Solution

1. Configure CacheModule.registerAsync() in AppModule with a Redis store and env-configurable TTL (CACHE_TTL)
2. Apply @UseInterceptors(CacheInterceptor) to:
   • GET /api/admin/dashboard (TTL: 60s)
   • Public menu listing endpoint (TTL: 300s)
   • Public restaurant info endpoint (TTL: 300s)
3. Use @CacheKey() with restaurant-scoped keys: restaurant:{id}:resource
4. Call cacheManager.del(key) in mutation service methods to invalidate on write

Acceptance Criteria

• Second identical GET request does not generate a new database query (verified via TypeORM query logging)
• Cache is invalidated after a menu item is created, updated, or deleted
• Redis key naming follows the restaurant:{id}:resource convention
• TTL configurable via CACHE_TTL env variable
• Cache miss falls through to database transparently