Preserve singularity of qform() and sform() individually

[tashrifbillah]

Refer to https://github.com/pnlbwh/dMRIharmonization/blob/spm-bspline/lib/resampling.py#L46

nibabel puts an "extra" 1 in row 4, column 4 when saving image:

(base) [tb571@pnl-z840-2 test_data_release]$ fslhd /tmp/resampled_mask.nii.gz | grep _xyz
ERROR: Illegal NIfTI file - /tmp/resampled_mask
       Zero determinant stored in sform and/or qform that is marked as valid
qto_xyz:1      -2.000000  -0.000000  -0.000000  110.588867
qto_xyz:2      0.000000  1.785172  -0.901755  -99.087921
qto_xyz:3      -0.000000  0.901755  1.785172  -96.036179
qto_xyz:4      0.000000  0.000000  0.000000  1.000000
sto_xyz:1      0.000000  0.000000  0.000000  0.000000
sto_xyz:2      0.000000  0.000000  0.000000  0.000000
sto_xyz:3      0.000000  0.000000  0.000000  0.000000
sto_xyz:4      0.000000  0.000000  0.000000  1.000000
# notice the 1 in (4,4) position although it was not present in the given sform()
# thus sform() becomes singular

(base) [tb571@pnl-z840-2 test_data_release]$ fslhd Site1_Y_mask.nii.gz | grep _xyz
qto_xyz:1      -2.000000  -0.000000  -0.000000  110.588867
qto_xyz:2      0.000000  1.785172  -0.901755  -99.087921
qto_xyz:3      -0.000000  0.901755  1.785172  -96.036179
qto_xyz:4      0.000000  0.000000  0.000000  1.000000
sto_xyz:1      0.000000  0.000000  0.000000  0.000000
sto_xyz:2      0.000000  0.000000  0.000000  0.000000
sto_xyz:3      0.000000  0.000000  0.000000  0.000000
sto_xyz:4      0.000000  0.000000  0.000000  0.000000
# in the given sform(), the matrix was completely zero, so no question about singularity/non-singularity

@suheyla2 's new test data masks have this "irregular" transform. Although they are irregular, we may be able to cover this base.

source: nibabel impact: minor TODO: devise a way to force the "extra" 1 to be zero, or make sform() and qform() completely non-singular.

[tashrifbillah]

PR #51