Bugfix s10: Multiple direct images per orbit for multiple visits

[K-Angelique-Kahle]

I currently have the pleasure to analyze data for multiple visits with multiple direct images per orbit. In s10, this triggers the "third case" in di_reformat in util.py.
The current version of pacman fails to correctly transfer the data stored in xrefyref.txt to meta.refpix, because the masking of orbits is not done correctly. To fix this, change line 267 in util.py from
mask_i = reffile['iorbit'] == i
to
mask_i = reffile['iorbit_cumul'] == i

I assume the current version will work if you only analyze one visit.

[sebastian-zieba]

Thanks Angelique! I'll make a PR for that.