test idempotence of trim_me in strings3.rs

[Rudxain]

Technically, this change doesn't directly test for idempotence, as the fn is called once instead of being chained. I still believe it adds value, because it can make the challenge "harder" while still being fair.

Note

The following is just feedback from my recent experience. It can be ignored as "off-topic", but I believe it might be helpful for designing exercises in general.

BTW, when I was solving it, I felt like using trim method was "cheating", so I implemented it myself:

fn trim_me(input: &str) -> &str {
    let l = |c| c != ' ';
    let start = input.find(l).unwrap_or_default();
    // it's easier to let patterns do their job
    // than to handle arithmetic overflows manually,
    // hence the asymmetry
    match input.rfind(l) {
        Some(end) => &input[start..=end],
        _ => &input[start..],
    }
}

Originally, the code was this:

fn trim_me(input: &str) -> &str {
    let l = |c| c != ' ';
    &input[input.find(l).unwrap_or_default()..=input.rfind(l).unwrap_or(input.len() - 1)]
}

Aside from being an eyesore, it doesn't work for "" (input.len() - 1 is always an invalid index). That's why I added some extra tests locally, as a "challenge" to myself.

Should I add the empty str test?