Nightqa v3

[araichoor]

This PR adds few modifications to the nightqa page creation.
The principal motivation is to move away from restricting to Main tiles only.
The chosen approach now considers all OBSTYPE="SCIENCE" exposures.

For the sframesky, tileqa, skyzfiber plots, we thus now include any SCIENCE observations.
The reason is because the type of survey should not prevent to inspect the exposures quality on those plots.

For the petalnz plot, we proceed a bit differently, as we stack the n(z) per tracer, per night.
We here:

• restrict to sv1, sv2, sv3, main (which conveniently have the same bits for the 5 plotted tracers);
• we make a serie of plots for each survey of the night;
• we still generate one file only.

Along the way, several changes - hopefully improvements:

• get_surveys_night_expids(): not using NTSSURVY anymore, we now rely on the fiberassign file which should be in the same folder as the desi-EXPID.fits.fz file
• write_nightqa_html(): look for possible nightlog filenames
• desi_night_qa: add a --steps argument to allow to execute only some steps
• protect against cases where no DARK exposure is found
• replace sys.exit(1) by some raise Errors

I've run it on what is currently processed in $DESI_ROOT/spectro/redux/f3:
https://data.desi.lbl.gov/desi/users/raichoor/nightqa_v3_dev/f3
(though note that those files may change if I do further developments/tests for this PR)

From a quick random spot checking of those: https://data.desi.lbl.gov/desi/users/raichoor/nightqa_v3_dev/f3/20210511/dark-20210511.pdf => r1 has very funny/dubious masked pixels distribution.

[coveralls]

Coverage decreased (-0.02%) to 25.637% when pulling c6d1fcf on nightqa_v3 into 8be844e on master.

[sbailey]

Looks good, thanks. I think you found a cosmic ray shower in r1. I put two minor comments in line, but ok to merge whenever you are ready.

[araichoor]

Good you waited to merge: I've noticed that I forgot to push my last commit last night: I've just done it (0c3bad2).
It's a simple "if enough tiles" case in the create_petalnz_pdf() function.
And I've also made the two small changes you noticed (thanks for the careful code reading!).

So I think I'm good on my side.