Optimize Regex match check

[HalidOdat]

No description provided.

[jedel1043]

@raskad This is a pretty big optimization because it removes O(n^2) complexity from our match implementation, but we are not sure how to adapt this to be able to work on unicode matching. Can you give this a look?

[github-actions]

Test262 conformance changes

Test result	main count	PR count	difference
Total	50,268	50,268	0
Passed	42,773	42,773	0
Ignored	1,391	1,391	0
Failed	6,104	6,104	0
Panics	18	18	0
Conformance	85.09%	85.09%	0.00%

[HalidOdat]

Reduced the failing tests from 48 to 4 unicode ones.

[HalidOdat]

After some more debugging I think it's an issue with regresses InputUtf16 iterator, managed to get a minimal testcase. We are trying to match against the lower part of the unicode 𝌆 this should not be possible in unicode mode since it's unicode aware. (In non-unicode mode it matches as expected).

/// 262 test/built-ins/RegExp/prototype/exec/u-lastindex-adv.js
///
/// Test case:
///
/// ```JavaScript
/// assert.sameValue(/\udf06/u.exec('\ud834\udf06'), null);
/// ```
#[test]
fn utf16_correct_unicode_scan() {
    // '𝌆' This is "Tetragram For Centre"
    // See: https://www.compart.com/en/unicode/U+1D306
    const INPUT: &[u16] = &[0xd834, 0xdf06];
    const MATCHER: &[u16] = &[0xdf06];

    let regex = Regex::from_unicode(MATCHER.iter().copied().map(u32::from), Flags::from("u"))
        .expect("valid regex");
    let m = regex.find_from_utf16(INPUT, 0).next();

    println!("{m:#?}");

    assert!(m.is_none());
}

There is a match:

Some(
    Match {
        range: 1..2,
        captures: [],
        named_captures: {},
    },
)

[raskad]

The tests seem fine with the new regress version so I think the only thing missing is the todo comment

[HalidOdat]

I put the comments this is ready for review/merge :)

[raskad]

Great find and optimization!