Revisiting Robust RAG: Do We Still Need Complex Robust Training in the Era of Powerful LLMs?