Map fsLR template to fsnative/T1w space

[oesteban]

In pursuing nipreps/fmriprep#1025, this would be the fastest route to avoid the overhead of an extra surface registration step with MSM. After all, users in need of a more precise registration should be using ciftify anyways.

To do this, I think this would be the workflow:

• Make sure we have all we need from templateflow (Ensure all necessary files are available under fsaverage and fsLR templateflow/templateflow#13)
• Back-project tpl-fsLR_space-fsaverage_hemi-{L,R}_den-32k_sphere.surf.gii to fsnative.
• Back-project tpl-fsLR_res-01_desc-aseg_dseg.nii.gz to T1w space.
• Project sub-NNN_space-T1w_desc-fsLRatlas_dseg.nii.gz to BOLD run space (this would happen on fmriprep proper)
• Sample via mris_vol2surf, as we are currently doing for GIfTI outputs (this would happen on fmriprep proper).

@edickie does this make sense?
@mgxd still interested on working this out or we should get this off your plate?
@effigies, any conceptual concerns about this?

[oesteban]

i don't think this is an issue anymore (see nipreps/fmriprep#1887 & nipreps/niworkflows#417)