Add first_modno input argument for components.JUNGFRAU

[kakhahmed]

This MR fixes issues with:

• Plotting xarray for FXE_XAD_JF500K which has one module JNGFR03.
• Overwrite and correcting modno_to_source and source_to_modno with the correct modno corresponding to the module array index.

This fix will used by https://git.xfel.eu/calibration/pycalibration/-/merge_requests/775 [Not yet tested with all reference runs.] It was tested with SPB_IRDA_JF4M and FXE_XAD_JF500K

And it is an alternative for European-XFEL/EXtra-geom#206

This issue was discovered after creating components.JUNGFRAU object for FXE_XAD_JF500K to form a DataArray for data.adc, then plotting it using plot_data_fast. The problem is that the module coordinates for this DataArray are of a value 3. As the data channel has JNGFR03.

Both modno_to_source and source_to_modno are modified when st_modno is set to > 0 value.
This helps in fixing the cases mentioned below in the comments with JF detectors with one module and receiver with integer > 1

Tested were added for st_modno and mock_fxe_jungfrau_run was added for /gpfs/exfel/exp/FXE/202101/p002478/raw/r0052

[kakhahmed]

I am not sure what the reason for the codecov/project failed check

[kakhahmed]

I am not sure anymore if this is the right approach.

There are more examples of this with other detector names.

I know that this is the karabo_da shown here in CALCAT but the detector's receiver sources are named based on it as far as I know.

[takluyver]

Ugh 🙄 . Do you know if this is only an issue with JF500K detectors? We could go back to your idea of applying it to all single module detectors. Or we could let the user pass in something like modnum_minus=2.

Or if the user specifies n_modules=4 or similar and we have 4 modules, we could automatically renumber them 1, 2, 3, 4. The only reason we can't do that without n_modules is that if we see only JNGFR02, we assume that it's a 2-module detector and the first module is missing from the data.

[kakhahmed]

It is with JF500K detectors for now. But there is no guarantee that this will be the case always.

Also, adding more info to this. The below screenshot shows that in offline calibration we are dependent on that DA/Receiver relation.

Or if the user specifies n_modules=4 or similar and we have 4 modules, we could automatically renumber them 1, 2, 3, 4. The only reason we can't do that without n_modules is that if we see only JNGFR02, we assume that it's a 2-module detector and the first module is missing from the data.

I always think about this solution. The only thing I couldn't solve is:

For example, we have JF4M -> JNGFR02 & JNGFR03 & JNGFR04 & JNGFR05.

The user passes n_modules = 4 and we see only JNGFR02 how would we know that this is the 1st detector?
Maybe we add a new argument for the first receiver name? e.g. first_receiver_name = "JNGFR02"

Passing n_modules and first_receiver_name might solve this case.

[takluyver]

The user passes n_modules = 4 and we see only JNGFR02 how would we know that this is the 1st detector? Maybe we add a new argument for the first receiver name? e.g. first_receiver_name = "JNGFR02"

Yeah, if there are missing modules, you'd have to specify where the numbers start from. I'd probably do it with passing an integer rather than a string, because this 'JNGFR02' part isn't something that EXtra-data otherwise asks you to look at.

[kakhahmed]

I'd probably do it with passing an integer rather than a string, because this 'JNGFR02' part isn't something that EXtra-data otherwise asks you to look at.

Yes, that makes sense 👍🏽

[kakhahmed]

I have replaced the changes in the MR by adding st_modno input argument to be able to point to the components.JUNGFRAU on the first modno for a JF detector.

[takluyver]

I think this looks like a good direction, my comments are all on the details.

Sorry it's taken so long to look at it again.

[takluyver]

Oh, and I meant to say, don't worry too much about codecov - it often seems to report that PRs cause a small drop in coverage. I imagine there's some fiddly issue with coverage collection, but so far I haven't bothered to figure it out.

[takluyver]

Thanks @kakhahmed !