[llvm] [LV][AArch64]: Utilise SVE ld4/st4 instructions via auto-vectorisation (PR #89018)

[Paul Walker via llvm-commits]

paulwalker-arm wrote:

> It loos like at the TargetLowering, LV and InterleavedAccessPass changes could be decoupled?

I agree.  Please can you split this PR into two PRs with the first only concerned with the correct lowering of the new emulated ld4/st4 IR sequence.  The second PR then teaches LoopVectorize how to use them.

https://github.com/llvm/llvm-project/pull/89018