i benchmarked it with hyperfine and in terms of performance it's pretty much identical, with a slight (fraction of a percent) advantage to my implementation
by setting `group_imports = "StdExternalCrate"` in `rustfmt.toml`, cargo fmt/rustfmt rearranges your use statements into three groups:
- std/core/alloc,
- followed by external crates,
- followed by self/super/crate