psImageCountPixelMask had counting errors for region tests

[eugene]

I fixed a variety of errors related to end-post interpretation of the psRegion
values. specifically, given an image with col0,row0 and numCols,numRows, a
region is valid if the upper limit is <= col0 + numCols. similarly, the value
of col1 should be col0 + numCols, not col0 + numCols - 1.

When I provided this code in rel10_ifa, I had used the psRegionForImage function
to rationalize the input region for the image. That was stripped out and
replaced with an idiosyncratic set of tests. I worry that these tests are not
all correct or consistent, which is why I used the rationalizing function. I
have added the fix to CVS HEAD

[Paul Price]

Fixed by EAM.