Only fuse pointwise with layernorm on the first argument

[pfultz2]

The kernel always assumes the first argument is the layernorm so we can't fuse it if its not the first argument. This undoes #2189 which didnt have a verification test in the first place.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.23%. Comparing base (64fe0c5) to head (22259a2).
Report is 2 commits behind head on develop.

Additional details and impacted files

@@           Coverage Diff            @@
##           develop    #3679   +/-   ##
========================================
  Coverage    92.23%   92.23%           
========================================
  Files          514      514           
  Lines        21746    21746           
========================================
  Hits         20057    20057           
  Misses        1689     1689           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[migraphx-bot]

Test	Batch	Rate new 22259a	Rate old a30b25	Diff	Compare
torchvision-resnet50	64	3,257.63	3,254.31	0.10%	✅
torchvision-resnet50_fp16	64	6,986.95	6,986.93	0.00%	✅
torchvision-densenet121	32	2,433.09	2,432.26	0.03%	✅
torchvision-densenet121_fp16	32	4,070.33	4,102.15	-0.78%	✅
torchvision-inceptionv3	32	1,628.57	1,627.22	0.08%	✅
torchvision-inceptionv3_fp16	32	2,748.08	2,744.59	0.13%	✅
cadene-inceptionv4	16	764.68	764.44	0.03%	✅
cadene-resnext64x4	16	813.64	812.75	0.11%	✅
slim-mobilenet	64	7,458.17	7,463.39	-0.07%	✅
slim-nasnetalarge	64	209.14	208.94	0.09%	✅
slim-resnet50v2	64	3,438.62	3,440.19	-0.05%	✅
bert-mrpc-onnx	8	1,151.96	1,143.08	0.78%	✅
bert-mrpc-tf	1	473.73	469.87	0.82%	✅
pytorch-examples-wlang-gru	1	419.85	513.89	-18.30%	🔴
pytorch-examples-wlang-lstm	1	393.31	386.92	1.65%	✅
torchvision-resnet50_1	1	816.53	776.87	5.10%	🔆
cadene-dpn92_1	1	398.12	395.29	0.72%	✅
cadene-resnext101_1	1	382.70	373.27	2.53%	✅
onnx-taau-downsample	1	345.54	345.34	0.06%	✅
dlrm-criteoterabyte	1	33.33	33.31	0.08%	✅
dlrm-criteoterabyte_fp16	1	52.51	52.72	-0.40%	✅
agentmodel	1	8,270.00	8,185.78	1.03%	✅
unet_fp16	2	58.84	58.75	0.16%	✅
resnet50v1_fp16	1	954.45	945.03	1.00%	✅
resnet50v1_int8	1	1,005.01	987.38	1.78%	✅
bert_base_cased_fp16	64	1,169.79	1,169.56	0.02%	✅
bert_large_uncased_fp16	32	363.63	363.05	0.16%	✅
bert_large_fp16	1	200.55	198.54	1.01%	✅
distilgpt2_fp16	16	2,197.94	2,199.18	-0.06%	✅
yolov5s	1	525.57	534.87	-1.74%	✅
tinyllama	1	43.36	43.39	-0.07%	✅
vicuna-fastchat	1	172.84	175.67	-1.61%	✅
whisper-tiny-encoder	1	416.57	417.72	-0.28%	✅
whisper-tiny-decoder	1	426.25	428.36	-0.49%	✅

This build is not recommended to merge 🔴

[migraphx-bot]

     ✅ bert-mrpc-onnx: PASSED: MIGraphX meets tolerance

     ✅ bert-mrpc-tf: PASSED: MIGraphX meets tolerance

     ✅ pytorch-examples-wlang-gru: PASSED: MIGraphX meets tolerance

     ✅ pytorch-examples-wlang-lstm: PASSED: MIGraphX meets tolerance

     ✅ torchvision-resnet50_1: PASSED: MIGraphX meets tolerance

     ✅ cadene-dpn92_1: PASSED: MIGraphX meets tolerance

     ✅ cadene-resnext101_1: PASSED: MIGraphX meets tolerance

     ✅ dlrm-criteoterabyte: PASSED: MIGraphX meets tolerance

     ✅ agentmodel: PASSED: MIGraphX meets tolerance

     ✅ unet: PASSED: MIGraphX meets tolerance

     ✅ resnet50v1: PASSED: MIGraphX meets tolerance

     ✅ bert_base_cased_fp16: PASSED: MIGraphX meets tolerance

🔴bert_large_uncased_fp16: FAILED: MIGraphX is not within tolerance - check verbose output

     ✅ bert_large: PASSED: MIGraphX meets tolerance

     ✅ yolov5s: PASSED: MIGraphX meets tolerance

     ✅ tinyllama: PASSED: MIGraphX meets tolerance

     ✅ vicuna-fastchat: PASSED: MIGraphX meets tolerance

     ✅ whisper-tiny-encoder: PASSED: MIGraphX meets tolerance

     ✅ whisper-tiny-decoder: PASSED: MIGraphX meets tolerance

     ✅ distilgpt2_fp16: PASSED: MIGraphX meets tolerance