There seems no term, people always need to describe that difference with saying things like 'global grid' vs. 'many, transformable volume objects', or whatever.
'True voxels' would be an idea, but it degrades the second option to be somewhat worse or half assed, which is not justified. (Why should we constantly re-voxelize our dynamic objects to a global grid just to display them? / Why should we accept the resulting temporal aliasing?)
I would only agree to your preference of global voxels if we talk about content creation but not game runtimes.
Well, after using voxels myself since some time i understand Siles volumetric vision a bit better. He's right with some things, and long before i was following his work with great interest.
But then they lost me with bold claims about a volumetric future, a paradigm shift, a revolution that makes it all better.
Maybe i would believe this if they would show some nice tools about volumetric content creation. But all i see is scanned content, and you can poke some spheres out of that. That does not really make content creation easier or possible at all.
Non the less, a demo game would be great. I would not mind if characters or vehicles use traditional triangles.