I'm a big fan of this project, and of adding backends generally, and of creating alternate compiler implementations even more generally. For those interested in the Binaryen route there's some work already in mir2wasm.
To the question at hand of cretonne, it seems to me there's an opportunity to make some powerful advances in the state of the art wrt high-performance code generation: the industry seems to be pretty content that LLVM is good enough, but we know it has some major limitations, primarily that it is slow and that it suffers from the undefined behavior spectre. If a competitive code generator appeared that didn't suffer these problems it could make a big splash. There's probably a lot of trepidation that the amount of work to catch up to LLVM would be massive, but we do have the advantage of writing cretonne in Rust, which is much easier to maintain and rely on than C++, and we have a viable incremental strategy of supporting debug mode x86 first and slowly competing with LLVM. Having a fast code generator is a critical step to solving slow compile times, which is Rust's Achilles heel.
I suspect this wouldn't be a major focus of the Rust team members this year, but hopefully there are others interested in making progress with our support (and cretonne development is ongoing regardless of what Rust is doing).