Adding Surface support

[romainVala]

🚀 Feature
For now we have Image Intensity and label data type but we are missing an other : surfaces
This is similar to label type where only spatial transform would be applied (not intensities transforms)

Motivation
An ultimate segmentation of brain tissue, should go to surface extraction...
I did not find a lot of work trying to predict cortical surface directly from a MRI 3D volume.
Actually just this one,
https://bitbucket.csiro.au/projects/CRCPMAX/repos/deepcsr/browse/README.md

Pitch

There are different step to will be need here :
_ Choose, the supported surface file format, and chose the library we need to read write surface files
_ Evaluate how to perform spatial transform (affine and elastic,
_ add some specific surface transform / sampler

Alternatives

stay with traditional segmentation...

Additional context

In deepcsr work he use the trimesh library
for instance here
https://bitbucket.csiro.au/projects/CRCPMAX/repos/deepcsr/browse/preprop.py#159

there is just one line, to apply an affine transform.
https://bitbucket.csiro.au/projects/CRCPMAX/repos/deepcsr/browse/preprop.py#162

Not sure how to apply elatic deformation here ... if possible

This is a new project we are starting, not how fare we will get, but I still open an issue, to start the discussion, and may be find other people interested in this ... and could help coding the solution