Hi Daniel,
thank you, adding:
-master_epi BASE
to align_epi_anat.py options solved the issue,
Just to see how it works I tried to go through 3dWarp too, but with no -gridset applied it cropped the T2 in the same way as in my original post.
However, providing the T1 as the gridset produced a virtually identical result to align_epi_anat.py with -master_epi BASE (minute differences I guess due to different interpolation method). So I could go either way though -master_epi BASE is obviously more convenient, no additional step as opposed to two additional steps (3dWarp precede by onverting the transform from 12x1 to 3x4 format).
Thanks again, and I apologize that - as I just noticed - I did not include my full align_epi_anat.py call in the first post. I intended to!