A few (admittedly negative) thoughts about automatic dimensioning:
Dimensions are subject to what I once heard called "secondary notation". That is, their usefulness and intelligibility is influenced by where they are placed in a drawing. In 2D, you have to deal with issues such as dimensions placing atop other dimensions or objects and clutter from too many redundant dimensions (when multiple parts are the same size). There is some art required to place dimensions for best use by humans. But until you modify the model they are ok once placed.
This situation is even more complicated in 3D than in 2D because the camera can orbit around the model interactively, even after everything is drawn. Like other model contents, dimensions can be obscured by other objects in front of them, can be seen edge on so they effectively vanish, and so on. As a result, 3D dimensions tend to be useful only after you have frozen a particular view, when you can choose their placement and orientation considering the rest of the view contents.
So my bottom line is skepticism: until someone shows me otherwise, I suspect it would take more effort to clean up automatically generated dimensions than it does for me to place them manually.