(First a question of Discourse etiquette - should I be replying to my own messages or should I be editing them?)
I have had a chance to look at bruker2nifti.py, and it does not deal with the orientation information correctly either - at least judged by whether my test phantom acquired in axial, coronal & sagittal orientations lines up on itself in FSLView.
As best as I can tell, there appears to be a bug in the 2dseq format for 2D coronal. I have now special-cased this in my ITK reader and my images all line up. I will test more thoroughly here and submit a patch on gerrit in the near future.