feat: Add simd shuffle / swizzle aliases for i16x8, i32x4, i64x2, f32x4 and f64x2

[MaxGraey]

• I've read the contributing guidelines
• I've added my name and email to the NOTICE file

[dcodeIO]

Are these instructions planned in Wasm? Otherwise it seems that these should not look like instructions but using the generic variant is preferred.

[MaxGraey]

No, it's just aliases. At the very least for shuffle I see no reason to have only the generic version, as it violates the general conventions. That is, I see no reason why we should only write v128.shuffle(....) instead of i16x8.shuffle but still have i8x16.shuffle available as alias v128.shuffle(...). This looks very inconsistent.

[dcodeIO]

I believe i8x16.shuffle is the (only) instruction existing in Wasm, motivating the generic variant (that also provides convenient fillers), while there is no i16x8.shuffle in Wasm. Thought the convention is rather not to add instructions that aren't already there respectively planned hmm.

[MaxGraey]

Well, in Wasm SIMD we have only one i8x16.shuffle but we have generic wrapper for specific scalar type. The main problem with generic wrapper is imprecise size of arguments (which is variable). Specialized aliases restrict signatures to a specific number of arguments, which provide much better experience. Also, code will look much more uniform and easier for refactoring. For example if user used i8x16.shuffle at first time and decided switched to i16x8 he needs refactor it to v128.shuffle<u16> instead i16x8.shuffle which may confused especially for first time