Superop fix

Created by: PhilipVinc

Closes #607 (closed)

This PR switches our representation of super-operators so stacked-rows (which is what you would have by doing np.kron(a,b), and the most natural for numpy/python since it is row-major).

I was previously using stacked-columns mainly because I was comparing to qutip (which is stacked-columns for historical reasons) and with my old Julia code (which is column major). The errors rised in #607 (closed) where due to the fact that stacked-columns required us to take a matrix transpose at some point.

This should make it more natural to work with our density matrices in the future, but it makes our representation incompatible with qutip, so we should add some conversion methods...