No reason to panic, submitted inadvertently!

Nice solution, but n > 0 can be omitted maybe?

I agree, sort_by_key is preferable. I honestly don't remember anything about this kata, but probably an unstable sort should also suffice? (Not in my solution obviously, but in the sort_by_key solution)

Because &name[0..1] is not a valid string if the first character spans multiple bytes.

The kata description is not very clear about the meaning of 'valid strings'.
Therefore, I would take it as 'any' string.
This solutions fails on the empty string as well as non-ASCII names like Иван

m is the maximum partition size. So sub(5, 3) calculates all partitions of 5 with no partition larger than 3.

I assume you mean that i % 2 returns -1 for negative odd numbers whereas i & 1 is positive. But I fail to see where behavior is different for even numbers. See https://play.rust-lang.org/?version=stable&mode=debug&edition=2018&gist=1ba4abb31460c86f5c426d188672c5bb

I think this solution might work for many problems. Isn't that kind of weird?!

It does a bitwise AND. As all odd numbers will have the final bit set and all even numbers not, this is a good test.

This is definitely an issue. As a workaround you can use the following template:

mod solution {
    pub use super::range_extraction;
}

pub fn range_extraction(a: &[i32]) -> String {
    // Your solution here
}