No doubt this would be a lot of fun. I'm not suggesting at all that it wouldn't be an awesome thing to start. I'm for Rust all the things!
I've wanted this myself, and for the specific use case I mentioned. The challenge I have is selling this project at work. I may have a reason to do it that involves implementing health monitors on the router for different nodes. This would be cool because of the ease of implementing state machines in Rust, it would be great for me.
The point I'm making is that if we can collect usecases, we can start working towards a common goal... I've decided to hold off until the hyper integration with tokio stabilizes before starting something like this.