[llvm] [AArch64] Avoid ld4 lowering for deinterleave4 feeding uitofp (PR #200763)

[Graham Hunter via llvm-commits]

huntergr-arm wrote:

FYI, the pass I'm trying to add in #193951 solves the same problem a bit earlier (before the InterleavedAccess pass), but only for scalable vectors right now. It could be extended to support fixed vectors. (There's also a reverse in the imagick loop that can be pulled into the tbl, which is planned as future work).

https://github.com/llvm/llvm-project/pull/200763