[webgpu] Always use tile matmulnbits for block_size = 32

[qjia7]

Description

After the optimization of prefill time with #23102, it seems that always using the tile matmulnibits with block_size = 32 can bring better performance even for discrete gpu for phi3 model.

Phi3 becomes 42.64 tokens/sec from 32.82 tokens/sec in easy mode on my NV RTX 2000 GPU.

[qjia7]

@guschmue @fs-eire Please help check other gpus you have at hand to see the overall results. Thanks.

[guschmue]

yes, I can confirm - I had taken out the check for intel and saw gains on a2000, 3060 and m4.
Can run a full benchmark later today.

[guschmue]

/azp run ONNX Runtime Web CI Pipeline,Windows GPU CI Pipeline,Linux Android Emulator QNN CI Pipeline

[guschmue]

/azp run Linux CPU CI Pipeline,Linux CPU Minimal Build E2E CI Pipeline,Linux GPU CI Pipeline,Linux GPU TensorRT CI Pipeline,Linux OpenVINO CI Pipeline,Linux QNN CI Pipeline,MacOS CI Pipeline,Windows ARM64 QNN CI Pipeline,Windows CPU CI Pipeline

[guschmue]

/azp run Windows GPU TensorRT CI Pipeline,onnxruntime-binary-size-checks-ci-pipeline,orttraining-linux-ci-pipeline,orttraining-linux-gpu-ci-pipeline,orttraining-ortmodule-distributed,Windows x64 QNN CI Pipeline,Big Models

[azure-pipelines]

Azure Pipelines successfully started running 2 pipeline(s).

[guschmue]

/azp run Windows GPU CUDA CI Pipeline,Windows GPU DML CI Pipeline,Windows GPU Doc Gen CI Pipeline

[azure-pipelines]

Azure Pipelines successfully started running 3 pipeline(s).

[azure-pipelines]

Azure Pipelines successfully started running 4 pipeline(s).

[azure-pipelines]

Azure Pipelines successfully started running 9 pipeline(s).

[guschmue]

for all models / scenarios for the gpu's impacted:

token/sec sppedup: avg ratio=1.19, >10% speedup=56.0%, >10% slowdown=4.0%, inside 10%=40.0%
prefill(500) speedup: sum avg ratio=2.20, >10% speedup=100.0%, >10% slowdown=0.0%, inside 10%=0.0%
prefill(1000) speedup: sum_long avg ratio=2.11, >10% speedup=100.0%, >10% slowdown=0.0%, inside 10%=0.0%