Use a distinct ToString implementation for u128 and i128

[GuillaumeGomez]

Part of #135543.

Follow-up of #136264.

When working on #142098, I realized that i128 and u128 could also benefit from a distinct ToString implementation so here it.

The last commit is just me realizing that I forgot to add the format tests for usize and isize.

Here is the bench comparison:

bench name	last nightly	with this PR	diff
bench_i128	29.25 ns/iter (+/- 0.66)	17.52 ns/iter (+/- 0.7)	-40.1%
bench_u128	34.06 ns/iter (+/- 0.21)	16.1 ns/iter (+/- 0.6)	-52.7%

I used this code to test:

#![feature(test)]

extern crate test;

use test::{Bencher, black_box};

#[inline(always)]
fn convert_to_string<T: ToString>(n: T) -> String {
    n.to_string()
}

macro_rules! decl_benches {
    ($($name:ident: $ty:ident,)+) => {
        $(
	    #[bench]
            fn $name(c: &mut Bencher) {
                c.iter(|| convert_to_string(black_box({ let nb: $ty = 20; nb })));
            }
	)+
    }
}

decl_benches! {
    bench_u128: u128,
    bench_i128: i128,
}

[rustbot]

r? @workingjubilee

rustbot has assigned @workingjubilee.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[bors]

☔ The latest upstream changes (presumably #136594) made this pull request unmergeable. Please resolve the merge conflicts.

[GuillaumeGomez]

Fixed merge conflicts.

[workingjubilee]

Having done code review of things that use Rust's feature(min_specialization), I now consider "specialization" a dirty word.

[GuillaumeGomez]

Noted, thanks for the title update then. :)

[tgross35]

If you've tested locally, is there any measurable perf/size impact here?

[tgross35]

Maybe pop this constant out so it can be reused between u128 and i128?

[GuillaumeGomez]

Done.

[GuillaumeGomez]

@tgross35 I added the comparison in the first comment with the code I use to get the performance comparison.