[daip] suspected bug in BPASS

[Eric Greisen]

I have worked on and solved the problem although I do not pretend to
understand quite why 31DEC05 was failing as it did or 31DEC03 working
(sort of) as it did.  The weights to the fitting program did not
change properly for different channels in 05 so it always thought 13
was flagged.  But it set the gain to 1,0 for flagged channels rather
than magic blanks.  The latter were the only ones that would be
interpolated correctly and (not) used in normalization.  Anyway, the
AOC version should now be okay and is available via the MNJ.

Eric Greisen