Paul,
thanks you for the detailed answer, this is much appreciated. A final question.
Say I am interested in extracting AFNI functional data that already has 3x3x3 mm voxels using 3dmaskdump in the neurological convention.
In this case, I would add the option
-nbox 3 3 3
to the 3dmaskdump script, correct? The extracted data would then correspond to the neurological or LPI-SPM convention, unless, of course, I would mess up something later in Python, But at least from the AFNI side of things, all should be clear.