Well the question is what effect will that have on the engine?
For example how will pixel round deal with it, or even flipping.
Anyway this is actually the one argument against using power of two images.Um no. I use power of 2 images. Problem being that you can't have a 16x16 square fit into a hole that is 16px wide because the way it calculates collisions.
So in order to get a 16x16 box into a 16x16 hole, I have to make it less than 16x16. In order to do this without it moving at higher resolutions, pixel rounding must be on, and it must be very slightly smaller than the size you need.
This is really more of an argument for a different collision system, but since that isn't going to happen I'd like to settle for the ability to edit the BBox vertices up to at least the 2nd decimal place. You can place them manually with the mouse between pixel so I don't see the harm in letting you do it via keyboard.