minFracMask is still too small

The resolution to PIPE2D-1413 was to set minFracMask to 0.25, but I think that's still too small. Here are some pfsMerged files with minFracMask=0.25
and minFracMask=0.95
I think that the latter are significantly better (I could have shown pfsArm files with the masked bits indicated). The noise still allows for the bad pixels (I hope!), so there's no need to label them bad.

An alternative is to ignore these mask bits when merging the spectra.