I'm going to offer a different explanation, but first, the standard one: Mozilla doesn't want to miss a paradigm shift, especially one where the only players currently in it are very proprietary. They learned their lesson with mobile.

One thing that many of us, myself included, are very excited about regarding the Servo/WebRender/Quantum projects is the ability for a fast, parallel web rendering engine to open new possibilities, such as how V8's JIT significantly pushed the envelope of what was possible. Now that JavaScript is acceptably fast, and we have WebAssembly for cases where control over memory layout is necessary, network speeds are the bottleneck of most applications. WebRender's big concept of rendering the entire web at "maximum FPS" is pretty awesome, but it's not pushing the envelope by itself, since framework and application developers have been treating "minimize DOM manipulation" as an important optimization. There's clearly potential with WebRender, but if a majority of used webpages are already rendering at an adequate framerate due to framework-level optimizations (and prior browser optimizations like Electrolysis and Stylo), it may be viewed as superfluous, despite potentially huge advantages.

I think the research in this direction is beneficial, and as a result, I'd like the Servo project to be allowed to continue developing cool stuff. As Servo won't replace Gecko in Firefox, it needs to try more challenging tasks to continue to push its envelope. WebRender has a ton of potential in the WebVR space, since a Web Components-based VR system would rely heavily on a performant DOM. Such a project would push Servo to continue development in preparation for a more functional future web.