Consider mentioning Manuel Drehwald's LLVM offload work in Ecosystem section

It's still very early days and I'm not sure if normal people can even try it out yet. But I find that approach interesting because if we can get it to work, from my understanding, it means we can share a large fraction of the work with LLVM's C/++/Fortran OpenMP 4 backends, and that gives us access to a much larger workforce than Rust-on-GPU has ever enjoyed so far. So I'm curious to see what it will lead us to.

As far as I know, the main way to track progress right now is to subscribe to this github issue.

cc @ZuseZ4 for corrections and additional details.

找到网站的 Ecosystem 部分，并查看链接的 Rust project-goals issue，以了解 Manuel Drehwald 的 LLVM offload 工作的当前详情。根据所引用的讨论确认状态和措辞，然后添加简洁、准确的说明及链接；Ecosystem 部分应清楚地反映这项工作，但不要夸大其成熟度。