[X86] matchBinaryPermuteShuffle - match to X86ISD::SHLD funnel shift patterns

[RKSimon]

Add matchShuffleAsVSHLD helper to recognise shuffle masks that can fold to funnel shifts

Fixes #145276

[carlsotr]

Should these returns of Shift * 8 have been Shift * ScalarSizeInBits? This appears incorrect for i32 shuffles.

define <8 x i32> @combine_vpshldq_v8i32_shuffle(<8 x i32> %a, <8 x i32> %b) {
  %r = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 1, i32 8, i32 3, i32 10, i32 5, i32 12, i32 7, i32 14>
  ret <8 x i32> %r
}

https://godbolt.org/z/h4G9nP6v3