subpriority overrides, take 2

[sbailey]

This PR replaces #736 which I will close after submitting this.

This provides tools for overriding SUBPRIORITY when generating target files, to work around the accidental SUBPRIORITY reassignments from fiberassign in May 2021:

• desitarget.subpriority.override_subpriority(targets, override) - override targets['SUBPRIORITY'] in-place, for those TARGETIDs that appear in the override table.
• desitarget.subpriority.get_fiberassign_subpriorities used to scrape fiberassign TARGETS HDUs to build override lists for dark, bright, and sky.
  • examples in /global/cscratch1/sd/sjbailey/desi/targets/subpriorities*.fits
  • this can be called from the command line but purposefully isn't put into the default desitarget installation path, e.g.

fadir=/global/cfs/cdirs/desi/target/fiberassign/tiles/trunk
python $DESITARGET/py/desitarget/subpriority.py -o . \
    -i $fadir/001/*.fits.gz $fadir/002/*.fits.gz \
       $fadir/020/*.fits.gz $fadir/021/*.fits.gz

• select_targets --dark-subpriorities ... --bright-subpriorities ... options to provide override lists
• select_skies --sky-subpriorities ... and supplement_skies --sky-subpriorities subpriorities-sky.fits options to provide sky SUBPRIORITY override lists
• more unit tests related to subpriority overrides and writing files

One change of behavior in write_targets et al:

• subpriority=True means set any SUBPRIORITY==0.0, but leave any SUBPRIOITY>0.0 alone (different than current master which overrides all or nothing)
• if you really want to override all, set input SUBPRIORITY==0.0 before calling write_targets(..., subpriority=True).

Examples in /global/cscratch1/sd/sjbailey/desi/targets/default and override/. I believe I have tested the following but would appreciate additional checks:

• targets appear in the same order regardless of override options
• targets that differ in SUBPRIORITY are in the override list
• targets with the same SUBPRIORITY are not in the override list
• TARGETIDs in the override list have the SUBPRIORITIES from that list

@araichoor @geordie666

[coveralls]

Coverage decreased (-0.04%) to 58.939% when pulling 7715f8b on subpriority2 into b2b3f21 on master.

[dstndstn]

This all looks reasonable to me