Create cuda tensor from CudaSlice

[coreylowman]

Similar to how we can create a cpu tensor from a Vec, it'd be nice to be able to pass in a separately allocated CudaSlice.

The new method of this would look like:

impl Cuda {
    pub fn tensor_from_buf<S: Shape, E: Dtype>(&self, buf: CudaSlice<E>, shape: S) -> Tensor<S, E, Self> {
       ...
    }
}

And can be added in src/tensor/cuda/allocate.rs

[DogPawHat]

It looks like Cuda::build_tensor mostly does this already?

[murluki]

Are there any issues in this project available for grabs?