Pass all questions to LLM and Q&A bot

[amindadgar]

Just query the RAG all the times and if informative answer found then return it to the user else use the answer from the LLM. (removing the detect_rag_question phase)