Empowering Large Language Model for Continual Video Question Answering with Collaborative Prompting