Understandable. I use DQS in my project, but there is only one character on screen using it :D I haven't tested scenarios with more than 2 DQS characters. Is it really that bad? In Unreal one can easily switch between different skinning methods via the deformer graph. You could even implement add runtime switches which react to the currently applied LOD.
It's not the heavy DQS in itself, but in this context it's an aggravating factor because, with hair strand-based groom already sacrificing the GPU, adding physics-enabled clothing and DQS on top of that makes the performance cost significant. Unfortunately, the blanket is very short in VR, and you always have to make design choices that can't satisfy every aspect. Yes, Unreal is much more convenient from that point of view, but unfortunately, it's not very well suited for VR performance.