BUG: Check volumetric and surface morphs

@britta-wstnr mentioned that the volumetric morphs might not (sufficiently) uniformly represent the data in the destination subject's space (e.g., voxel 1 might show up 10% as much as voxel 2). This is something that I think we can and should investigate by at least comparing surface to volumetric morphs. In particular we can:

1. Compare the morph_mat sparse representations in terms of row and column norms (assuming we roughly match the numbers of rows and columns)
2. Compare the distribution of neighbor-distances in the original subject space (dirac delta at pos for volume, some smooth narrow-ish distribution around some dist in mm for surface) to those of the spatially equivalent points (probably based on argmax?) in the destination subject's space.

From these we can see if surface and volume have similar before/after distributions, so we at least know how comparable the two solutions are.

I have some adult data I could use for this, including with functional localizers.

Start by reviewing the volumetric and surface morph implementations that produce the morph_mat sparse representations and use pos for spatial points. Compare row and column norms, then compare neighbor-distance distributions between original and destination subject spaces using the adult data mentioned. Done means documenting whether the surface and volume morphs have comparable before-and-after distributions.