Includes part of the header propagates some calibration input files

I was looking at a header from `matterhorn/calibnight/20210412/psfnight-z3-20210412.fits and noticed in the `inc` part that there was mention of `everest`:

```
COMMENT inc /global/cfs/cdirs/desi/spectro/redux/everest/calibnight/20210621/psf
COMMENT night-z3-20210621.fits
COMMENT inc /global/cfs/cdirs/desi/spectro/redux/everest/calibnight/20210624/psf
COMMENT night-z3-20210624.fits
COMMENT inc /global/cfs/cdirs/desi/spectro/redux/everest/calibnight/20210626/psf
COMMENT night-z3-20210626.fits
COMMENT inc /global/cfs/cdirs/desi/spectro/redux/everest/calibnight/20210627/psf
COMMENT night-z3-20210627.fits
COMMENT inc /global/cfs/cdirs/desi/spectro/redux/everest/calibnight/20210628/psf
COMMENT night-z3-20210628.fits
COMMENT inc /global/cfs/cdirs/desi/spectro/redux/everest/calibnight/20210629/psf
COMMENT night-z3-20210629.fits
COMMENT inc /dvs_ro/cfs/cdirs/desi/spectro/redux/matterhorn/exposures/20210412/0
COMMENT 0084461/fit-psf-z3-00084461.fits
COMMENT inc /dvs_ro/cfs/cdirs/desi/spectro/redux/matterhorn/exposures/20210412/0
COMMENT 0084462/fit-psf-z3-00084462.fits
COMMENT inc /dvs_ro/cfs/cdirs/desi/spectro/redux/matterhorn/exposures/20210412/0
COMMENT 0084463/fit-psf-z3-00084463.fits
COMMENT inc /dvs_ro/cfs/cdirs/desi/spectro/redux/matterhorn/exposures/20210412/0
COMMENT 0084464/fit-psf-z3-00084464.fits
COMMENT inc /dvs_ro/cfs/cdirs/desi/spectro/redux/matterhorn/exposures/20210412/0
COMMENT 0084465/fit-psf-z3-00084465.fits
```

Talking about it with @akremin for a bit, I think these files are probably coming from the input `fibernight` from `desi_spectro_calib` as this is from the header:
`DEPNAM07= 'CCD_CALIB_PSF'
DEPVER07='SPCALIB/spec/sm6/psfmean-z3-20210601-20210630.fits' `

I also see this behavior in 20260405 (`/global/cfs/cdirs/desi/spectro/redux/matterhorn/calibnight/20260405`):

```
COMMENT inc /dvs_ro/cfs/cdirs/desi/spectro/redux/daily/exposures/20260201/003349
COMMENT 25/fit-psf-z3-00334925.fits
COMMENT inc /dvs_ro/cfs/cdirs/desi/spectro/redux/daily/exposures/20260201/003349
COMMENT 26/fit-psf-z3-00334926.fits
COMMENT inc /dvs_ro/cfs/cdirs/desi/spectro/redux/matterhorn/exposures/20260405/0
COMMENT 0345442/fit-psf-z3-00345442.fits
COMMENT inc /dvs_ro/cfs/cdirs/desi/spectro/redux/matterhorn/exposures/20260405/0
COMMENT 0345443/fit-psf-z3-00345443.fits
COMMENT inc /dvs_ro/cfs/cdirs/desi/spectro/redux/matterhorn/exposures/20260405/0
COMMENT 0345444/fit-psf-z3-00345444.fits
COMMENT inc /dvs_ro/cfs/cdirs/desi/spectro/redux/matterhorn/exposures/20260405/0
COMMENT 0345445/fit-psf-z3-00345445.fits
COMMENT inc /dvs_ro/cfs/cdirs/desi/spectro/redux/matterhorn/exposures/20260405/0
COMMENT 0345446/fit-psf-z3-00345446.fits
```

I am wondering if this in intended behavior or something we should cut as this info is in the `desi_spectro_calib` `fiber night`?

Start by tracing how desi_spectro_calib produces the fibernight input and how its CCD_CALIB_PSF dependency is written into the calibration FITS header. Compare the 20210412 and 20260405 examples, then determine whether the propagated inc entries are intended; done means documenting the decision and, if unintended, identifying the source change and validating headers without those entries.

Written by the indexing model from the issue text.