The original 8-corner bounds transform was necessary after all,
so this reverted the "savings" of only doing 2 corners and
put the origianl 8-corner logic back for parts' bounds.

Also, Vessel bounds was always using 2-corner logic and was
wrong all along - this is adding the 8-corner logic to it too.

Special thanks to @RCrockford in kOS discord for warning me
about this when I discussed my plan, thus making me check
again with a more complete test to catch that my idea didn't
really cover all situations (for the 2 corner method to fail
you have to be oriented a specific way relative to the axes,
and on the runway at KSC just so happened to NOT trigger that
case so it wasn't clear that it was wrong till I went into
orbit and tried again there where I could spin all axes.