Oh that is pretty neat, I'm still failing to come up with any current use cases though. Most animations I'm use to are bone based which I think would still be smaller than this. I could see it for vertex based animations, but I'm not sure of any current cases where this is used. I'm sure there are good uses out there, I'm probably just missing it. I can at least think of potential future uses like vr movies, where if it's recorded and reconstructed into a 3d scene the bone information won't be there and most likely won't help even if it was.

Most animation in VFX is exported as alembic or USD and generally the skinning/bone deformations are baked down to vertex animations and sent on to the fx, lighting and compositing departments in that simplified but data heavier form. Better compression could save us many terabytes where I work, and who knows how many petabytes across the industry. Most studios use some amount of cloud rendering and the reduction in network traffic and storage with those farms would be appreciated.