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

[Simon Pilgrim via llvm-commits]

RKSimon wrote:

```ll
; ModuleID = 'bugpoint-reduced-simplified.bc'
source_filename = "bug.ll"
target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-i128:128-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"

define void @blam(ptr readonly align 8 captures(none) dereferenceable(64) %arg) #0 {
  %getelementptr = getelementptr inbounds nuw i8, ptr %arg, i64 8
  %i = load <6 x i64>, ptr %getelementptr, align 8
  %i1 = shufflevector <6 x i64> %i, <6 x i64> poison, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
  store <4 x i64> %i1, ptr poison, align 8
  ret void
}
attributes #0 = { "target-features"="+avx" }
```

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