Add interactive test notebooks for open systems

[SeitherJulian]

Same as #43 but for open systems. Work in progress.

[ajgpitch]

As #51 now merged, hopefully this will pass tests

[pmenczel]

The notebooks are pretty much finished, however, there are still failing tests.

• For state transfer with CRAB (open system), the manually computed fidelity slightly differs from the one returned by QOC
• For gate synthesis problems in open systems with all algorithms, the manually computed fidelities are much worse than the ones returned by QOC. This might be due to unclear definition of fidelity in this case (I believe QOC / qtrl are choosing fidelity functions somewhat arbitrarily).
  But inspecting the actual Qobj, I find that the differences between the computed quantum channel and the target channel are huge, even if QOC reports very high fidelities.

(This is mostly notes for myself, to be discussed with @ajgpitch @nwlambert @rochisha0 at a future meeting)