Use simd intrinsics for max and min

[Nugine]

Copied from wasm32 implementaions

stdarch/crates/core_arch/src/wasm32/simd128.rs

Lines 2543 to 2554 in 547e3b0

	/// Compares lane-wise unsigned integers, and returns the maximum of
	/// each pair.
	#[inline]
	#[cfg_attr(test, assert_instr(i8x16.max_u))]
	#[target_feature(enable = "simd128")]
	#[doc(alias("i8x16.max_u"))]
	#[stable(feature = "wasm_simd", since = "1.54.0")]
	pub fn u8x16_max(a: v128, b: v128) -> v128 {
	let a = a.as_u8x16();
	let b = b.as_u8x16();
	unsafe { simd_select::<simd::i8x16, _>(simd_gt(a, b), a, b).v128() }
	}

Affected functions:

sse2:

• _mm_max_epi16
• _mm_max_epu8
• _mm_min_epi16
• _mm_min_epu8

sse41:

• _mm_max_epi8
• _mm_max_epu16
• _mm_max_epi32
• _mm_max_epu32
• _mm_min_epi8
• _mm_min_epu16
• _mm_min_epi32
• _mm_min_epu32

avx2:

• _mm256_max_epi16
• _mm256_max_epi32
• _mm256_max_epi8
• _mm256_max_epu16
• _mm256_max_epu32
• _mm256_max_epu8
• _mm256_min_epi16
• _mm256_min_epi32
• _mm256_min_epi8
• _mm256_min_epu16
• _mm256_min_epu32
• _mm256_min_epu8

[rust-highfive]

r? @Amanieu

(rust-highfive has picked a reviewer for you, use r? to override)