fix preproc/fibermap keyword propagation

[sbailey]

Header keyword updates: this PR fixes #963 (FIBERMAP missing OBSTYPE) and fixes #964 (inconsistent EXPTIME in different HDUs). Reviews are welcome if you can get them in by Friday noon, otherwise I'd like to merge Friday afternoon for test runs over the weekend prior to the NERSC outage next week.

In the case of EXPTIME, the problem was that the fibermap is generated from several input files and it was getting its EXPTIME from the guider file instead of the spectrograph raw data file, resulting in inconsistent numbers.

For OBSTYPE, it was originally coming from the guider file but arcs and flats don't have that input. Now the fibermap also gets keywords from the raw data file, and thus picks up OBSTYPE too.

Tested with:

assemble_fibermap -n 20200315 -e 55612 -o fibermap-00055612.fits
desi_preproc -n 20200315 -e 55612 --fibermap fibermap-00055612.fits \
    --cameras b0 --outdir .
desi_preproc -n 20200315 -e 55717 --cameras b0 --outdir .
fitsheader -k OBSTYPE -k EXPTIME preproc-b0-00055612.fits
fitsheader -k OBSTYPE -k EXPTIME preproc-b0-00055717.fits

Main branch:

# HDU 0 in preproc-b0-00055612.fits:
EXPTIME =                 900. / [s] Actual exposure time                       
# HDU 1 in preproc-b0-00055612.fits:
# HDU 2 in preproc-b0-00055612.fits:
# HDU 3 in preproc-b0-00055612.fits:
# HDU 4 in preproc-b0-00055612.fits:
OBSTYPE = 'SCIENCE '                                                            
EXPTIME =                 20.0                     

# HDU 0 in preproc-b0-00055717.fits:
EXPTIME =                  60. / [s] Actual exposure time                       
# HDU 1 in preproc-b0-00055717.fits:
# HDU 2 in preproc-b0-00055717.fits:
# HDU 3 in preproc-b0-00055717.fits:
# HDU 4 in preproc-b0-00055717.fits:                             

Note that OBSTYPE is missing from HDU 0, sometimes missing from HDU 4 (FIBERMAP) and EXPTIME is inconsistent between HDUs 0 and 4.

This PR:

# HDU 0 in preproc-b0-00055612.fits:
OBSTYPE = 'SCIENCE '                                                            
EXPTIME =                 900. / [s] Actual exposure time                       
# HDU 1 in preproc-b0-00055612.fits:
# HDU 2 in preproc-b0-00055612.fits:
# HDU 3 in preproc-b0-00055612.fits:
# HDU 4 in preproc-b0-00055612.fits:
OBSTYPE = 'SCIENCE '                                                            
EXPTIME =                900.0                                                  

# HDU 0 in preproc-b0-00055717.fits:
OBSTYPE = 'FLAT    '                                                            
EXPTIME =                  60. / [s] Actual exposure time                       
# HDU 1 in preproc-b0-00055717.fits:
# HDU 2 in preproc-b0-00055717.fits:
# HDU 3 in preproc-b0-00055717.fits:
# HDU 4 in preproc-b0-00055717.fits:
OBSTYPE = 'FLAT    '                                                            
EXPTIME =                 60.0                                                  

[julienguy]

Thanks for this. Checked on TESTARC data. Good to have the obstype keyword in the preprocessed arc lamp data.