Add utility function ansi::slice_ansi_str

[remi-dupre]

Hi!

While working on a bug of indicatif (cf. console-rs/indicatif#627), I needed an ANSI-aware slicing method. I think it made more sense to add it in this crate.

/// Slice a `&str` in terms of text width. This means that only the text
/// columns strictly between `start` and `stop` will be kept.
///
/// If a multi-columns character overlaps with the end of the interval it will
/// not be included. In such a case, the result will be less than `end - start`
/// columns wide.
///
/// If non-empty head and tail are specified, they are inserted between the
/// ANSI symbols from truncated bounds and the slice.
///
/// This ensures that escape codes are not screwed up in the process.
pub fn slice_str<'a>(s: &'a str, head: &str, bounds: Range<usize>, tail: &str) -> Cow<'a, str> {
    ...
}

I made this quite similar to truncate_str, which enables code reuse 😊

I also suggested a non-allocating implementation for measure_text_width 🤷

[remi-dupre]

This was just a random (new?) clippy lint : these are already u16 🤷

[remi-dupre]

Not related to the MR, but while implementing a less optimized draft of my fix I thought this could spare a few allocs. This is probably not something to care about so I wouldn't mind if you want to revert this for the sake of simplification.

[djc]

Can you avoid let .. else in order to pass the MSRV tests?

[remi-dupre]

Indeed, that should be good now 👍

[remi-dupre]

And format is also a recent-is feature, on it 😬

I'll try to build on 1.56, it didn't work last time I tried it because it fails to resolve regex = "=1.10.3" 😞

EDIT : Ok, got it

[remi-dupre]

Should be good, make test runs with rust 1.56 on my machine 👍