Update notebooks based on successful CI runs

[fingolfin]

We plan to provide (most/all?) of our examples as "static" notebooks which are pre-rendered, so people can easily view them. When we do that, it's annoying if the output in there does not match the current output -- e.g. if the OSCAR banner says "0.1" when the actual version is "0.5".

I am not actually sure what we test there right now: if the output differs, do we reject this as a test failure? Or do we just test that the example runs through and produces no error?

Regardless of that, we need an easy and quick way to update the notebooks with the "latest" outputs, and the logical thing to do here is to use the CI for this: i.e., as part of the CI, update the notebooks (either by pushing directly to their repos, or by opening PRs for them; both of these are not too hard to script)

[rbehrends]

We are currently strictly testing for errors occuring during tests, but don't compare output. Comparing them is not hard per se; I already render them to RST as a diff-friendly format, but we also need to have an easy way to update them if they change. I think the best way to go about that would be to put them in a separate repository, e.g. oscar-ci-data, which could also hold other such information.

[fingolfin]

@rbehrends already made some nice progress on this, see e.g. https://oscarci.mathematik.uni-kl.de/job/oscar/9126/artifact/logs/build-9126/g-vectors.ipynb.log