Rearrange CI tests

[teutoburg]

• Mark whole test class if all tests in that class were marked.
• Integrate (cloned) notebook tests as a step in the overall test suite, as they were triggered by the same events anyway and it's übersichtlicher this way.
• Use separate webtests and also "updated-dependencies" test from DevOps.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (3d174ef) 75.15% compared to head (c4d45c4) 77.04%.

Additional details and impacted files

@@              Coverage Diff               @@
##           dev_master     #341      +/-   ##
==============================================
+ Coverage       75.15%   77.04%   +1.88%     
==============================================
  Files              57       57              
  Lines            7693     7693              
==============================================
+ Hits             5782     5927     +145     
+ Misses           1911     1766     -145     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[hugobuddel]

I personally might prefer to have the notebook tests in a separate workflow from the regular tests, because we might decide to run them in different ways later.

In particular, I was hoping we could get the coverage of the regular tests up, so it would not be necessary to run the notebook tests as often. E.g. run the regular tests for each PR, but the notebooks only nightly. And maybe we should not run the notebooks on github at all, but on zeus in Vienna.

Those considerations apply mostly to the IRDB, and less so to ScopeSIM. As in, I suppose we do want to test the ScopeSim notebooks more often. So separating the tests here would be mainly to keep the repository structures similar.

Nevertheless/therefore, it is fine to make this change. It is indeed overzichtelijker this way.

[teutoburg]

I personally might prefer to have the notebook tests in a separate workflow from the regular tests, because we might decide to run them in different ways later.

In particular, I was hoping we could get the coverage of the regular tests up, so it would not be necessary to run the notebook tests as often. E.g. run the regular tests for each PR, but the notebooks only nightly. And maybe we should not run the notebooks on github at all, but on zeus in Vienna.

Those considerations apply mostly to the IRDB, and less so to ScopeSIM. As in, I suppose we do want to test the ScopeSim notebooks more often. So separating the tests here would be mainly to keep the repository structures similar.

Nevertheless/therefore, it is fine to make this change. It is indeed overzichtelijker this way.

Yeah, the IRDB is different in several ways, might come around to that in the near future...