Computational efficiency for surface operations

[akhanf]

The recent (merged and open) PRs with surface processing (#357, #342, and #359) have some computational efficiency issues, e.g. with surface generation now taking on order of minutes, and laplace-beltrami nearly an hour depending on surface complexity, we should revisit this code and try to use libraries (with vtk/c++ backends) instead of processing meshes directly with python/numpy.. pyvista would be the first choice to use since it is already in our envs, but there are likely other options as well..

[Dhananjhay]

@mackenziesnyder and I are looking into this.