Fix order of tensor operations in dtw_cpu call

[mishazawa]

convert to f64 after moving to CPU. This will work on MPS device (apple m1)

original error:

    return dtw_cpu(x.double().cpu().numpy())
                   ~~~~~~~~^^
TypeError: Cannot convert a MPS Tensor to float64 dtype as the MPS framework doesn't support float64. 
Please use float32 instead.