OCT Conversion Orientation Issue

[calvinchai]

The only question I have, and I think we'll have to check with Nate, is how the channels of the input file map to its spatial axes. Say differently, each voxel contains an orientation vector and we need to understnad in what referential this orientation vector is. In the output zarr, we'll want the first channel to correspond to dimension "x", the second channel to dimension "y", etc.

Originally posted by @balbasty in #19 (comment)