Orientation woes

[timholy]

In #11 I noticed an orientation mismatch, and according to http://teem.sourceforge.net/nrrd/format.html#general.4 Fiji is right and we are wrong. Gack. Another breaking change, coming up...

[timholy]

Ah, wait. That example is just illustrative, and it's valid to use a different ordering. But it's such a pervasive assumption that I suspect we should follow it in the absence of evidence to the contrary. The main negative is that if we return a lazy-permuted array (necessary if we're using mmapped files), then operations on these arrays will generally not be very cache-friendly.