adjust default psf wavelength solution if arcs fail

[sbailey]

Currently in the pipeline if we can't fit PSFs for at least 3 exposures on a night, we fall back to using the default PSF which has a potentially different wavelength solution. However, sometimes the nightly data isn't good enough to fit the full PSF shape but it is still good enough to fit the wavelength solution, which is still better than using the default PSFs as-is. Even if we don't have any arcs on that night, we could still adjust the default PSF wavelength solution using the first science exposure of the night, before proceeding with processing the flats (if we have flats).

This requires some delicate bookkeeping, so make sure to cover all cases of completely missing arcs, having arcs for some exposures but not enough, having arcs for some cameras but not others, missing arcs but having flats, missing both arcs and flats, etc.

The main point is that by adjusting the wavelength solution of the default PSF by using data from the current night, we can do better than either

• using both the default PSF and default flats
• using a non-wavelength adjusted default PSF to extract nightly flats