The data was little different than usual - the 4mm3 voxels are a little large, and the obliquity doesn't bring the dataset very close together, so I suspect that information is wrong in the header. However, I did get something to work reasonably well. Try this command with your data.
align_epi_anat.py -giant_move -anat subj01.anat+orig -epi epi_subj01.r01+orig. -giant_move \
-suffix _al2epi -deoblique off -epi_strip None -epi_base 0 -cost lpc+ZZ