Perform a conversational search

Performs a search query with conversational RAG enabled. Typesense retrieves relevant documents and passes them as context to the configured LLM, which generates a natural language answer. Supports multi-turn conversations by passing a conversation_id to maintain context across queries.