loongarch: Use unified data types for SIMD intrinsics

[heiher]

Suggested by @sayantn

[rustbot]

r? @sayantn

rustbot has assigned @sayantn.
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

[heiher]

Although type aliases won't break the build, they will trigger clippy wanrings. If the user's CI treats warnings as errors, this could cause the CI to fail.

warning: transmute from a type (`std::arch::loongarch64::m256i`) to itself
   --> src/simd/lines_bwd.rs:204:47
    |
204 |                 let tmp = lasx_xvpermi_q::<1>(T(sum), T(sum));
    |                                               ^^^^^^
    |
    = help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#useless_transmute
    = note: `#[warn(clippy::useless_transmute)]` on by default

[sayantn]

seeing that it is a clippy warning, not even rustc-builtin, I would say this is acceptable

[folkertdev]

re. the transmute warning, that is fine, but it's probably nicest for users to #[allow] the lint (or even #[expect] if you can make that work)

[heiher]

a gentle ping.

[folkertdev]

The technical changes look good to me, I've left some notes on the docs. Those are user-facing, so I've been a bit nit-picky there.

[folkertdev]

r? @folkertdev