[llvm] [WebAssembly] Lower fmuladd to madd and nmadd (PR #161355)

[Sam Parker via llvm-commits]

sparker-arm wrote:

Results still look good:

```
Metric                    Execution Time Change(%)
----------------------  --------------------------
Mean (filtered)                             14.854
Median (filtered)                           16.472
Max (filtered)                              25.636
Min (filtered)                               0.015
Geomean (non-filtered)                      13.953 

Benchmark                             Execution Time Change(%)
----------------------------------  --------------------------
ncnn-FastestDet-run_times                               11.111
ncnn-alexnet-run_times                                  16.683
ncnn-blazeface-run_times                                12.803
ncnn-efficientnet_b0-run_times                          11.927
ncnn-efficientnetv2_b0-run_times                        17.4
ncnn-googlenet-run_times                                18.773
ncnn-googlenet_int8-run_times                            0.627
ncnn-mnasnet-run_times                                  17.153
ncnn-mobilenet-run_times                                25.044
ncnn-mobilenet_ssd-run_times                            24.454
ncnn-mobilenet_ssd_int8-run_times                        0.015
ncnn-mobilenet_v2-run_times                             19.737
ncnn-mobilenet_v3-run_times                             14.865
ncnn-mobilenet_yolo-run_times                           25.636
ncnn-mobilenetv2_yolov3-run_times                       21.615
ncnn-nanodet_m-run_times                                18.885
ncnn-proxylessnasnet-run_times                          14.983
ncnn-regnety_400m-run_times                             16.26
ncnn-resnet18-run_times                                 19.614
ncnn-resnet50-run_times                                 23.153
ncnn-shufflenet-run_times                               15.25
ncnn-shufflenet_v2-run_times                            20.588
ncnn-squeezenet-run_times                               19.472
ncnn-squeezenet_int8-run_times                           0.451
ncnn-squeezenet_ssd-run_times                           13.072
ncnn-squeezenet_ssd_int8-run_times                       0.027
ncnn-vgg16-run_times                                    18.597
ncnn-vgg16_int8-run_times                                0.132
ncnn-vision_transformer-run_times                       13.76
ncnn-yolo_fastest_1.1-run_times                         12.267
ncnn-yolo_fastestv2-run_times                           12.088
ncnn-yolov4_tiny-run_times                              18.895
```


https://github.com/llvm/llvm-project/pull/161355