SSM Meets Video Diffusion Models: Efficient Video Generation with Structured State Spaces