possible bandgap.pl bug

[cjobrien]

I was using the bandgap.pl script when I noticed that, although the script will read the occupations from the last ionic step, it will use the occupancies from the first electronic step. This can pose problems as this value is not a converged solution; in fact, I found this behavior when the code was giving me errors about negative occupancy of bands and I attempted to verify this error by inspecting the OUTCAR file.