Quality report refactor

[mmccrackan]

Reworks quality reporting code to improve h5 storage and add NEP plots.

[kmharrington]

are these nets or neps?

[mmccrackan]

They're now NEPs, though I guess since this is inconsistent in a few places we should just rename the Influx fields to something generic like noise.

[mmccrackan]

Okay, I think I've incorporated discussions from the previous site-pipeline call into this. Compared to the version currently running on site, this PR implements the following:

• Overhaul saving of the data so that things like arrays aren't saved as strings. Do not save the entire PWV array, just the values for each obs_id.
• Exposes some of the main config options to the main functions so they can be configured in Prefect.
• Longterm dataset now accumulates when monthly reports are run.
• Add NEP vs PWV and NEP vs Time plots.
• Add obsdb value vs time plot. Currently have boresight angle, hwp speed, and corotator angle. Will add scan params and PWV when those become available in the regular obsdb.
• Added a coverage plot. Since this is slow, this is parallelized in the same way as preprocess_tod. It will take ~10 mins for the SATs and maybe 1-2 hours for the LAT on 4 processes. It saves a FITS file and a png file (the latter is what is plotted).
• Rearranged the plots into separate rows so everything mostly lines up in time.
• More stats at the top.
• Added a "weekly/latest" and "monthly/latest" to the drop down box at the top.
• Raises an error when any reports fail to prevent silent failures.

Example here: https://site.simonsobs.org/~mmccrackan/qr/satp3/weekly/20260107_20260114/report.html