[llvm] [WebAssembly] Add support for avgr_u in loops (PR #153252)

[Jasmine Tang via llvm-commits]

================

----------------
badumbatish wrote:

It doesn't seem to generate a avgr_u

```
  %a = add <16 x i16> %x, %y
  %b = add <16 x i16> %a, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1,
                          i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
  %c = lshr <16 x i16> %b, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1,
                           i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
  ret <16 x i16> %c
```
seems to generate this
```
; SIMD128-LABEL: avgr_u_v16i16:
; SIMD128:         .functype avgr_u_v16i16 (i32, v128, v128, v128, v128) -> ()
; SIMD128-NEXT:  # %bb.0:
; SIMD128-NEXT:    i16x8.add $push0=, $2, $4
; SIMD128-NEXT:    v128.const $push9=, 1, 1, 1, 1, 1, 1, 1, 1
; SIMD128-NEXT:    local.tee $push8=, $4=, $pop9
; SIMD128-NEXT:    i16x8.add $push1=, $pop0, $pop8
; SIMD128-NEXT:    i32.const $push2=, 1
; SIMD128-NEXT:    i16x8.shr_u $push3=, $pop1, $pop2
; SIMD128-NEXT:    v128.store 16($0), $pop3
; SIMD128-NEXT:    i16x8.add $push4=, $1, $3
; SIMD128-NEXT:    i16x8.add $push5=, $pop4, $4
; SIMD128-NEXT:    i32.const $push7=, 1
; SIMD128-NEXT:    i16x8.shr_u $push6=, $pop5, $pop7
; SIMD128-NEXT:    v128.store 0($0), $pop6
; SIMD128-NEXT:    return
;
```


https://github.com/llvm/llvm-project/pull/153252