This is as much a "showstopper" as the infamous "chkdsk" bug
It's worse because the cursor bug really exists while the chkdsk thing was part misunderstanding and part 3rd party chipset driver bug.
i.e. The chkdsk thing is 0 bad and the cursor thing is 0.00000001 bad.
Nobody is (seriously) calling it a showstopper, of course. Just a small cosmetic glitch that will drive some people with OCD crazy. (I can see it really easily but, thankfully, it doesn't bother me. It does have me curious about what the cause is, though,
as the .ani cursor file plays fine in other things.)
I solved the mystery. The Win7 busy-cursor files have two sizes embedded (48x48 and 32x32). The larger size is fine and are what you'll see if you open the file in most editors. Not sure if any ani-cursor tools understand the dual-size files so I used a
hex editor to extract the individual 32x32 frames. Those are the ones Windows actually uses (at default DPI anyway). It's then
easy to see that frame 6 is an exact copy of frame 7, and what should be frame 6 is missing, causing the glitch. (You can also see that the 48x48 versions are fine.)
Very simple explanation in the end. It only seemed confusing because the tools we were using to look at the file hid the fact it had two sizes of images inside of it.