`CudaSlice` equivalent of `Slice::split_at_mut`

[dkales]

The CudaSlice abstraction does not have an API to split it into two mutable CudaViewMuts, which kind of forces a user to do so with CudaViews, and passing references to those to kernels if they want to separate a CudaSlice into two sub-slices, which kind of defeats the point of trying to more cleanly work with CudaViewMuts.

It would be nice if there was an analogue to the std library function split_at_mut, for both CudaSlices and CudaViewMuts.