desi_run_night pernight redshift broken; internal_id bookkeeping

[sbailey]

desi_run_night -n 20230612 --proc-obstypes science --tiles 24782 --dry-run-level 1

fails with

INFO:processing.py:1597:submit_redshifts: Submitting joint redshift fits of type pernight for TILEID 24782.
INFO:processing.py:1648:submit_redshifts: Expids: [185028].

Traceback (most recent call last):
  File "/global/common/software/desi/users/sjbailey/desispec2/bin/desi_run_night", line 123, in <module>
    sys.exit(submit_night(**args.__dict__))
  File "/global/common/software/desi/users/sjbailey/desispec2/py/desispec/scripts/submit_night.py", line 511, in submit_night
    = submit_tilenight_and_redshifts(ptable, sciences, calibjobs, internal_id,
  File "/global/common/software/desi/users/sjbailey/desispec2/py/desispec/workflow/processing.py", line 2095, in submit_tilenight_and_redshifts
    ptable, internal_id = submit_redshifts(ptable, sciences, tnight, internal_id,
  File "/global/common/software/desi/users/sjbailey/desispec2/py/desispec/workflow/processing.py", line 1649, in submit_redshifts
    redshift_prow, internal_id = make_redshift_prow(zprows, tnight, descriptor=zsubtype, internal_id=internal_id)
ValueError: too many values to unpack (expected 2)

It looks like make_redshift_prow returns only the redshift_prow, not the internal_id, but it is also taking the internal_id as an input. It looks like on lines 1601-1602,internal_id is incremented outside of make_redshift_prow, while make_joint_prow increments the internal_id inside the function and returns the new one to use. Let's make that consistent across functions.

We don't necessarily need pernight redshifts for kibo, so I'll pass this to @akremin to inspect rather than making a last minute emergency patch. If this fix is easy, we could incorporate it before the final tag to launch kibo this afternoon.

[akremin]

This was caused by an incomplete refactor in a previous pull request, which slipped through the cracks. I originally planned to shift from the use of make_redshift_prow to make_joint_row, but didn't quite make the transition here. I have reverted back to using make_redshift_prow as was already implemented. That required removing interal_id as an expected output and incrementing it manually. This is solved in PR #2339 .

I will also point out that we should be using desi_proc_night rather than desi_run_night. To reduce confusion I have also removed desi_run_night in the referenced PR. That should no longer be needed since desi_proc_night has successfully replaced it for many months now. I had originally left it so that it could be removed in a single "house cleaning" pull request, but it now appears that won't happen before Kibo and it is useful to remove this confusion from the tagged desispec.

[akremin]

Resolved with PR #2339