[Bug] split_long_word incorrectly splits multi-codepoint character

[uasi]

Describe the bug

split_long_word() may split multi-codepoint characters, such as â (a + ◌̂ in NFD) and 🇺🇳, in the middle.

Steps to reproduce

use comfy_table::utils::formatting::content_split::split_long_word;
use unicode_width::UnicodeWidthStr;

fn main() {
    let emoji = "🙂‍↕️"; // U+1F642 U+200D U+2195 U+FE0F head shaking vertically
    assert_eq!(emoji.len(), 13);
    assert_eq!(emoji.chars().count(), 4);
    assert_eq!(emoji.width(), 2);

    let (word, remaining) = split_long_word(emoji.width(), &emoji);

    println!("{word}");
    assert_eq!(word, "\u{1F642}\u{200D}");
    assert_eq!(word.len(), 7);
    assert_eq!(word.chars().count(), 2);
    assert_eq!(word.width(), 2);

    println!("{remaining}");
    assert_eq!(remaining, "\u{2195}\u{FE0F}");
    assert_eq!(remaining.len(), 6);
    assert_eq!(remaining.chars().count(), 2);
    assert_eq!(remaining.width(), 2);
}

Logs (if applicable)

No response

Operating system

macOS 15.2

Comfy-table version

v7.1.3

Additional context

No response

[tisonkun]

unicode-rs provides a library to handle clusterng - https://github.com/unicode-rs/unicode-segmentation

Maybe we can see if it helps.

[Nukesor]

Patch is released :)