Replace aspect_repository with rustc_env for exposing workspace_name

[UebelAndre]

This is minor cleanup I've found to be easier to understand and work with in comparison to aspect_repository.

[hlopko]

This change will break us internally (for a surprising reason - we don't support external repositories, so using @ labels like @rules_rust doesn't work). Since you describe it as a minor cleanup, and since the aspect_repository only exists temporarily because of bazelbuild/bazel#11734, I'm curious if we could not submit this PR and wait a bit longer for the Bazel issue to be fixed. Then we can delete the workaround.

I've pinged the Bazel issue again now.

CC @krasimirgg

[UebelAndre]

This change will break us internally (for a surprising reason - we don't support external repositories, so using @ labels like @rules_rust doesn't work). Since you describe it as a minor cleanup, and since the aspect_repository only exists temporarily because of bazelbuild/bazel#11734, I'm curious if we could not submit this PR and wait a bit longer for the Bazel issue to be fixed. Then we can delete the workaround.

I've pinged the Bazel issue again now.

CC @krasimirgg

That issue is only relevant to the rust_analyzer target. rustfmt still needs to know the workspace name and had actually stopped using consuming the variable set to aspect_repository. So there it'd be a no-op cleanup but figured the same mechanism could be used to solve for this issue and reduce the footprint introduced by bazelbuild/bazel#11734 by making another rule slightly more generic and moving comments over to the only logic impacted by that. At the very least I'd like to keep the cleanup for rustfmt but would want to do some tighter colocation of this code in the mean time. I don't expect any fix for an open Bazel issue to make it's way into rules_rust any time soon.

[UebelAndre]

On second thought, it current feels like negative value to generalize the workspace name stuff. I'll decline this PR for now but it's quite a useful trick for finding runfiles. Maybe it should be built into the runfiles library?