The images are not overscan subtracted nor trimmed since there is no standard trimsec or biassec
A lot of the bad pixels and some columns cancel out in the division of one flatfield by another
The main bad columns are detected from the ratio image
Picks up the main bad columns (including the one on the left
which is not seen by the previous BADPIX mask)
The faint column seen at the right of the FLATHIGH image cancels
out in the divided image (FLATLOWHIGH) and is not picked up in the
The two other bad columns at the top right have almost
disappeared in the divided image and hence are also not picked up in
the BADPIX mask
Picks up the main bad columns from the divided image (including
the one on the left which is not seen by the previous BADPIX mask)
The "middle" bad column has almost disappeared in the divided
image and hence is not masked in the BADPIX mask.
The edge of the FOV (seen in the FLATHIGH.fits and the previous
BADPIX mask) is not evident in the BADPIX mask made by mkBADPIX.pl
This is not important though, since this area will have to be trimmed
in the reduction process anyway.
Picks up the main bad column on the left hand side, but not the faint column seen in FLATHIGH.fits since this divides out in the divided image.