[AGIPD][DARK] Avoid failed jobs for creating constants and injecting them

Description

This is a preparation for integrating the calibration for AGIPD single modules.

At the moment dark processing will have only one failed job while creating the dark summary plots.

Now summary plotting is skipped by default for single module detector.

How Has This Been Tested?

Relevant Documents (optional)

Types of changes

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

Checklist:

Reviewers

@schmidtp @hammerd

added Doing label

assigned to @ahmedk

added 1 commit

• 64f7c66d - remove dinstance and use karabo_id to decide the correct plot

Compare with previous version

added 1 commit

• 5fbc06fd - remove dinstance and use karabo_id to decide the correct plot

Compare with previous version

added 1 commit

• d2b72d55 - remove dinstance and use karabo_id to decide the correct plot

Compare with previous version

added 1 commit

• 3a426252 - Propagate show_processed_modules change to LPD and DSSC and summary notebook

Compare with previous version

Two small comments, otherwise fine. It was soul-crushing to see all that hardcoded strings in there, thankfully they can go away soon.

added 1 commit

• 9303a724 - remove the dinstance leftovers and use nmods instead of checking 1M and 500K

Compare with previous version

added 1 commit

• 8914b9dc - move module_shape assignmet up

Compare with previous version

resolved all threads

resolved all threads

resolved all threads

Thanks, LGTM!

This doesn't even look like much of a workaround anymore

True. It's not.

But it's not the final change needed for a single module. I am skipping here the summary plot. next step is making it work properly.

Thank you for the review. Are we planning to deploy this soon?

Ah yes, I forgot about that. Can we disable the summary plot somehow? A while ago for the HED ONC case, we enabled returning from a notebook early via sys.exit(0).

I'm checking back with Silo, but it does not seem urgent immediately. Just always nice to be prepared.

For correction, we can skip_plots, yes.

For dark there is a way to skip it but not directly. If local constants aren't available local_output == False, the Summary notebook exits and is skipped.

In this MR I print a message and skip, anyway even if local_output == True, if we are trying to plot a single module AGIPD detector.

Oh, you do. I'm sorry, I understand your point now, I was confused by the MR description where you're still mentioning a single failed job. But it simply doesn't plot for now.

I checked EXtra-geom, and indeed it looks like we need to create a custom detector (i.e. inherit from GenericGeometry or DetectorGeometryBase) with only a single module.

Ops, sorry for the confusion. I aimed a bit low at the beginning of this MR.

changed the description

added 1 commit

• f5761a62 - move skipping notebook up

Compare with previous version

Okay, final LGTM. But data is now available, let's have a look.

mentioned in merge request !977 (merged)