[TVM EP][CI] Integrate TVM EP into ORT public CI on Windows

[KJlaccHoeUM9l]

Description
In our previous PR#11851 we added the ability to use TVM EP on Windows.
This PR integrates tests for checking TVM EP on Windows into public ORT CI.

Notes about LLVM
TVM EP requires LLVM. LLVM is already installed in Windows 2019 image we use to run the pipeline. However, this package is not suitable for building TVM EP. This is due to the fact that the packages of official LLVM releases for Windows (from their official GitHub) contain only runtime components, but for TVM we also need to have compilation libraries that are in LLVM-Dev.

This issue has been raised many times before:

• LLVM/Clang Windows installer doesn't provide CMake package information llvm/llvm-project#47222 (comment)
• llvm installer on windows missing LLVMConfig.cmake llvm/llvm-project#53052
• LLVM installer for Windows does not provide llvm-config llvm/llvm-project#35139
• https://www.reddit.com/r/cpp_questions/comments/d986yu/llvmconfig_on_windows/

To solve this problem on Windows, it is proposed to build LLVM yourself from the source code.

However, there is a faster way -- use Miniconda to install LLVM-Dev. And after installation, expand PATH environment variable with the following value: <miniconda_dir>\Library\bin.

[KJlaccHoeUM9l]

Hello @xadupre!
Could you review the PR?
Also, an error occurred with Lint Python, but it is not related to this PR. The same error is observed in another PR: https://github.com/microsoft/onnxruntime/runs/7319325550?check_suite_focus=true

[KJlaccHoeUM9l]

The problem seems to be on the pyright side: microsoft/pyright#3689.

[KJlaccHoeUM9l]

Hello @xadupre!
The problem with pyright has been resolved: microsoft/pyright@879ab89
Now can we continue the review and run the remaining pipelines?

[xadupre]

/azp run Windows CPU CI Pipeline, Windows GPU CI Pipeline, Windows GPU TensorRT CI Pipeline, Windows WebAssembly CI Pipeline, orttraining-amd-gpu-ci-pipeline, orttraining-linux-ci-pipeline, orttraining-linux-gpu-ci-pipeline, orttraining-ortmodule-distributed, onnxruntime-python-checks-ci-pipeline

[azure-pipelines]

Azure Pipelines successfully started running 8 pipeline(s).

[xadupre]

/azp run Linux CPU CI Pipeline, Linux CPU Minimal Build E2E CI Pipeline, Linux GPU CI Pipeline, Linux GPU TensorRT CI Pipeline, Linux Nuphar CI Pipeline, Linux OpenVINO CI Pipeline, MacOS CI Pipeline, ONNX Runtime Web CI Pipeline, onnxruntime-binary-size-checks-ci-pipeline, Linux TVM CI Pipeline

[azure-pipelines]

Azure Pipelines successfully started running 9 pipeline(s).

[xadupre]

Sorry for the delay, I'm off today and the next day.

[KJlaccHoeUM9l]

@xadupre, ok, come back when you're available.
All tests passed successfully. If you have no comments, can we merge the PR?

[xadupre]

Any particular reason you chose this version of python?

[KJlaccHoeUM9l]

On Ubuntu 20.04 this version comes out of the box. This version was the most tested (during development).
This version is also used in the Linux pipeline.
СС @snnn

[xadupre]

Should I merge?

[KJlaccHoeUM9l]

@xadupre,
yes, the PR is ready to merge.