ENH: insert gradunwarp in bold workflow

[bpinsard]

Changes proposed in this pull request

The goal of that PR is to integrate gradient unwarping to the BOLD preprocessing workflows (and to anat workflows through update of smriprep).
See #1550.
Relies on an updated version of the gradunwarp https://github.com/bpinsard/gradunwarp/tree/fix/mem_leaks.

Unwarping is applied on the initial ref (to use a single volume, it mainly takes into account the matrix size and affine rather than image content).
Warp-field is then passed to the bold_bold, bold_T1w, bold_std one-step resamplings.

Still a WIP: inserting gradient unwarping seems more challenging for BOLD than for anatomical images. I need to check how that integrates with sdcflows.

Necessary PRs:
nipreps/smriprep#355
nipreps/niworkflows#819

Documentation that should be reviewed

TODO: document the option to pass a gradient coeff file.

[codecov]

Codecov Report

Patch coverage: 36.36% and project coverage change: -0.14% ⚠️

Comparison is base (e1c4bac) 49.29% compared to head (71845e5) 49.15%.
Report is 8 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #3064      +/-   ##
==========================================
- Coverage   49.29%   49.15%   -0.14%     
==========================================
  Files          46       46              
  Lines        3593     3609      +16     
==========================================
+ Hits         1771     1774       +3     
- Misses       1822     1835      +13     

Files Changed	Coverage Δ
fmriprep/workflows/base.py	10.00% <ø> (-0.26%)	⬇️
fmriprep/workflows/bold/base.py	19.84% <0.00%> (-0.71%)	⬇️
fmriprep/workflows/bold/registration.py	10.63% <ø> (ø)
fmriprep/workflows/bold/resampling.py	11.20% <ø> (ø)
fmriprep/cli/parser.py	82.74% <100.00%> (+0.07%)	⬆️
fmriprep/config.py	82.08% <100.00%> (+0.08%)	⬆️

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.