Add a marching cube transform to get surface representations

Is your feature request related to a problem? Please describe.
This is not related to a problem but I wonder why there is not such a transform as a wrapper for marching cube function implemented in scikit-image.

Describe the solution you'd like
As I found that some transforms as KeepLargestConnectedComponent require installation of scikit-image, I wonder why a function that transfom binary segmentations to points and vertex does not exist in the library.

Drawback : such a transform could not be composed with other transforms. And I'm not sure it has a meaning to extend it to more that 3D datas.

Describe alternatives you've considered
Using directly the marching cube implementation of scikit-image on my transformed data (instance of MetaTensor)
vertices, triangles, _, _ = ski.measure.marching_cubes(data.numpy(), 0.5)

Start by reviewing the scikit-image ski.measure.marching_cubes entry point and the MetaTensor usage shown in the issue. Clarify the intended MONAI transform interface, especially whether it should support only 3D binary segmentations and how its vertices and triangles are returned. Done means the requested surface representation is available through a documented transform with appropriate composition behavior.