This is only true under the assumption that you can analyse the video and extract the movement/features as quickly as you can stream video to the network. Given how optimised streaming of live video is and how hard video analysis can be, it's not at all a given.
I'm not sure I follow? When you use Workrooms you're not streaming video.
The application is taking the positional data of your headset/controllers and translating that into how your avatar moves around. There isn't any video analysis going on.
You said "the only visual data that needs to be sent across network is the positioning of facial features and hand movements". To extract facial features and hand movements is non-trivial and requires video analysis and may hence involve lag.
