[llvm] [AMDGPU] Fix folding of v2i16/v2f16 splat imms (PR #72709)

[Jay Foad via llvm-commits]

jayfoad wrote:

> We can use inline constants with packed 16-bit operands, but these should use op_sel.

The docs say that there are special rules for this for f16/bf16 dot2 instructions - see the RDNA3 Instruction Set Architecture
Reference Guide section 7.2.1. "Non-Standard Uses of Operand Fields" subsection "Inline constants with DOT2_F16_F16 and DOT2_BF16_BF16".

So does this code need to treat f16/bf16 dot2 instructions differently from other packed instructions?

https://github.com/llvm/llvm-project/pull/72709