Fix Flaky Tests

[Lukas113]

Since two months, our scheduled tests don't succeed anymore on a regular basis. However, to me it seems like the tests since the last three weeks work or fail, depending on how the stars are aligned. I'll just list the errors which are flaky and seem to not be fixed yet (since three weeks):

• test_parallelization_by_observation -> dirty.fits.pybdsf.log not found, is happening since 3 weeks (see run)
• test_HIIM_Img_Recovery_notebook -> reason: unknown (see run), just happened once yesterday

These bugs (especially the first one) needs to be fixed ASAP. They also affect active PR (their tests fail).

[Lukas113]

I guessed it might have something to do with different installed versions of packages in the environment. However, looking at the diffs of the according runs I don't think so anymore. I haven't found any common denominator.

[Lukas113]

The tests don't fail anymore since 1 1/2 months (see CI-Test 880 on 01.12.2023). According to the error message (see image below), I'm pretty sure that the work on the Karabo-Feedstock at the 1. Dec fixed this issue, because since then, the boost-versions of pybdsf are fixed (as it should be according to the conda-doc) and we have pretty much a pybdsf build for each boost-version available.