Make fast kronecker matvec behavior more intuitive

[nchristensen]

The existing behavior was based on a Matlab script, but is unintuitive for Python. (I think it has something to do with Matlab using F ordering.) The order of the arguments should also by Sx, Sy, Sz rather than Sz, Sy, Sx to match the standard mathematical notation $S_x \otimes S_y \otimes S_z$

[nchristensen]

We should probably also change the function name to avoid collision with Scipy's and Numpy's kron.