[llvm] [X86] Remove LowerFCanonicalize and use generic expansion (PR #147877)

[via llvm-commits]

woruyu wrote:

> Do not add a new TLI hook for this

The hook is used to distinguish between X86 and (AMDGPU systemZ) behavior for vector FCANONICALIZE.
X86 needs to directly call expandFCanonicalizeWithStrictFmul, otherwise UnrollVectorOp produces poor code.
In contrast, AMDGPU requires UnrollVectorOp; bypassing it leads to instruction selection failures.
Any suggestion for this, I don't konw how to deal with it.

https://github.com/llvm/llvm-project/pull/147877