[AArch64] Suboptimal `memmove` for non-power of two sizes

[Kmeakin]

Split off from #165625

https://godbolt.org/z/TafacnzY3

LLVM assembly:

move7:
        ldrb    w8, [x1, #6]
        ldrh    w9, [x1, #4]
        ldr     w10, [x1]
        strb    w8, [x0, #6]
        strh    w9, [x0, #4]
        str     w10, [x0]
        ret

GCC assembly:

move7:
        ldr     w2, [x1]
        ldr     w1, [x1, 3]
        str     w2, [x0]
        str     w1, [x0, 3]
        ret

[llvmbot]

@llvm/issue-subscribers-backend-aarch64

Author: Karl Meakin (Kmeakin)

Split off from https://github.com//issues/165625

https://godbolt.org/z/TafacnzY3

LLVM assembly:

move7:
        ldrb    w8, [x1, #<!-- -->6]
        ldrh    w9, [x1, #<!-- -->4]
        ldr     w10, [x1]
        strb    w8, [x0, #<!-- -->6]
        strh    w9, [x0, #<!-- -->4]
        str     w10, [x0]
        ret

GCC assembly:

move7:
        ldr     w2, [x1]
        ldr     w1, [x1, 3]
        str     w2, [x0]
        str     w1, [x0, 3]
        ret