[llvm] [X86] combineConcatVectorOps - concat per-lane v2f64/v4f64 shuffles into vXf64 vshufpd (PR #143017)

[Alexander Kornienko via llvm-commits]

alexfh wrote:

And a simpler reproducer (using code after SLP vectorizer): https://gcc.godbolt.org/z/bMGrr8Gze

https://github.com/llvm/llvm-project/pull/143017