[daip] IMAGR ver. 31DEC07 does not work for ARCH=SUL

[Eric Greisen]

Oh dear - you have found a problem.  I wonder if something is working
because the usual offset is negative on linux and is positive here.  I
did find just such a bug already but that should be fixed.  That was
QPTDIV which would not affect IMAGR - doing bad things to CALIB
mostly.  The original dirty image seems fine but the inner loop
subtraction must be bad since it goes immediately to NITER and says
the max residual is essentially the original max.  The final residual
and image are bizarre too.

Thanks - I need this sort of help and will look into it.

Eric Greisen