SPMamba: State-space model is all you need in speech separation