delay calibration - workflow with logic to automatically get the best one

[tikk3r]

Continues on https://github.com/LOFAR-VLBI/lofar-vlbi-pipeline/wiki/2024-11-28 which determined a metric for selecting a handful or promising delay calibrator candidates with logic to find the actual best one based on phasediff scores used for selecting DD calibrators.

[tikk3r]

https://git.astron.nl/RD/VLBI-cwl/-/tree/auto-di-selection?ref_type=heads

[tikk3r]

First iteration implemented with https://git.astron.nl/RD/VLBI-cwl/-/merge_requests/88, which selects the user-specified N best sources and calibrates them against user-supplied skymodels.

[tikk3r]

Phasediff score versus current pipeline metric ("g score") for all LBCS candidates in LockmanC:

Best g score

image 0	image 1	image 9

Best phasediff score

image 0	image 1	image 9

The lowest phasediff selected the more compact source (which is the better source, but would benefit from LOFAR-VLBI/VLBI-cwl#18 due to high dynamic range).

[lmorabit]

depends on automatic DD calibration workflow(s) from Jurjen - so just waiting for that to be merged.

[jurjen93]

The DD calibration workflow has already been merged last month. So, this can go ahead, right?

[tikk3r]

The DD calibration workflow has already been merged last month. So, this can go ahead, right?

Putting the slack discussion here for reference: it still needs review comments and other small things address. It would also benefit from another discussion about whether to combine it with a rewrite of some core pipeline logic to simplify the code base, or to leave that for a separate ticket (see the comments on the GitLab MR for context).

[tikk3r]

I think implementing this logic will be quite powerful in combination with the automatic settings. So my proposal is we implement the functionality, leaving the bigger simplification refactor for a refinement ticket.