[LPD] [Dark] Don't use multiprocessing inside multiprocessing when opening run

Description

There was an error opening the run if its JSON cache file didn't already exist, because it tried to use multiprocessing inside multiprocessing, which is not allowed. This avoids that scenario. Behaviour should be unchanged if the cache file exists.

How Has This Been Tested?

No specific tests, but the parallelize=False option was added for precisely this use case - where you're already using a multiprocessing pool outside opening the run - so I'm confident in it.

Types of changes

• Bug fix (non-breaking change which fixes an issue)

Checklist:

• My code follows the code style of this project.

Reviewers

@schmidtp @ahmedk

LGTM

LGTM

mentioned in merge request !879 (merged)

added 2 commits

• f0aa8385 - return for modules with 0 trains
• b85b0d65 - Merge branch 'fix/avoid_errors_for_empty_modules' into 'fix/lpd-dark-daemon-children'

Compare with previous version

added 2 commits

• c405d64b - fix: accept uppercase calibration type
• 58b4d429 - Merge branch 'fix/accept_uppercase_calibration' into 'fix/lpd-dark-daemon-children'

Compare with previous version

Uhm, is it intentional some other fixes have been merged into this one?

changed milestone to %3.11.0

The CI was failing; I guess Karim wanted to fix that before merging this to master.

Yes, The CI didnt need to be merged on this one. But I guess this happened because it was rebased on this MR. The other fix MR was intentionally merged onto it. I expected that we merge this MR first before the others.

And yes as @kluyvert explained, the CI was failing. But it has succeeded so no need to run it again https://git.xfel.eu/calibration/pycalibration/-/jobs/446933

merged

mentioned in commit 02cf43fc