Non-power-of-two sized volumes do not resample correctly

This might be related to #47.

I noticed when looking at some newly acquired scan data that sections of the scans were completely missing in the volume view. On a hunch, I padded the data to a power of two dimension on the one axis where it wasn't yet and \*poof\*, suddenly all my data was visible.

I'm still unsure what's going on here, but this should be enough information to start a deeper investigation.