Sounds like more of a CYA thing than anything. It's obvious that you are right in that the USB power ends up on the +5V rail, but it appears that happens from within the mega32u usb chip? At any rate, +5V is being applied to the output of the regulator and not the input when powering from USB. The datasheet makes it clear that this shouldn't cause a problem with the regulator either. I suspect that maybe these "regulated" supplies provided an initial unregulated spike at power on.