[BUG] 2-Qubit Kraus Error Channel

[stroblme]

As a result of #71 which is (partially) closed with #72, we should re-visit the implementation of noise channels for the 2-qubit case (see ansaetze.py).
There is a very informative discussion here (thanks @majafranz for spotting this) and the follow-up post here.
However, implementation seems not trivial, but the current way of dealing with it is also not correct.