Semantic Surface could be a class instead of dict?

[balazsdukai]

Currently its a bit confusing that there is a Geometry class with its methods and such, but the Semantic Surfaces are just plain dict-s. Thus when getting the boundaries of the sem surf-s, we need to use the Geometry methods, and need to know that we are accessing dictionaries. Not very intuitive.

groundsrf = geom.get_surfaces(type='groundsurface')
ground_boundaries = [geom.get_surface_boundaries(g) for g in groundsrf.values()]

I think I would prefer sth like below. Although, it would add quite a bit of overhead for keeping all those semantic objects, esp. in triangulated models.

groundsrf = geom.get_surfaces(type='groundsurface')
ground_boundaries = [srf.boundary for srf in groundsrf]