✨ (memory): implement semantic search for memory retrieval

Introduce semantic search capabilities to enhance memory retrieval
based on conversation context. Add vector embedding support to the
Memory data structure and integrate OpenAIEmbeddings for generating
embeddings. Implement a new function, `generate_loading_query`, to
create search queries from conversation states. Update memory storage
and retrieval processes to utilize vector embeddings, enabling more
accurate and relevant memory recall. This enhancement improves the
agent's ability to provide contextually relevant responses by
leveraging past interactions more effectively.