Implementing LLM Response Caching with Redis: Semantic Deduplication, TTL Strategies, and Cache Invalidation Patterns | Wicked Smart Data