Update the quantization script to support GEMM (transB==1)

[flans39]

Description:

• Update replace_gemm_with_matmul to support quantization of GEMM nodes whose transB equals 1.

Motivation and Context

• Currently, the onnxruntime quantization script can convert a GEMM node to MatMulInteger if the transB attribute of the node equals 0. However, in many onnx models (e.g. https://github.com/onnx/models/tree/master/vision/classification/vgg), the transB attribute of the GEMM nodes is 1. Thus, these nodes are not able to be quantized by the current script.
• This PR fixes a closed issue (expand to support ‘gemm’ quantization #2340).

[ghost]

All CLA requirements met.

[kit1980]

/azp run Linux CPU CI Pipeline, Linux CPU x64 NoContribops CI Pipeline, Linux GPU CI Pipeline, Linux GPU TensorRT CI Pipeline, Linux OpenVINO CI Pipeline

[azure-pipelines]

Azure Pipelines successfully started running 5 pipeline(s).

[yufenglee]

Thanks for contributing! Could you please add the handling of the case that B is not an initializer?

[yufenglee]

/azp run Linux CPU CI Pipeline, Linux CPU x64 NoContribops CI Pipeline, Linux GPU CI Pipeline, Linux GPU TensorRT CI Pipeline, Linux OpenVINO CI Pipeline,Linux CPU Minimal Build E2E CI Pipeline

[azure-pipelines]

Azure Pipelines successfully started running 5 pipeline(s).

[yufenglee]

/azp run orttraining-linux-ci-pipeline,orttraining-mac-ci-pipeline,Linux OpenVINO CI Pipeline,Windows GPU TensorRT CI Pipeline, centos7_cpu, orttraining-linux-gpu-ci-pipeline

[azure-pipelines]

Azure Pipelines successfully started running 6 pipeline(s).

[yufenglee]

/azp run Windows GPU CI Pipeline,Windows CPU CI Pipeline,MacOS NoContribops CI Pipeline,MacOS CI Pipeline,Linux CPU CI Pipeline

[azure-pipelines]

Azure Pipelines successfully started running 4 pipeline(s).

[yufenglee]

/azp run Linux CPU CI Pipeline