[Feature]: Enforce formatting standards for C++ and CUDA code

[mgoin]

🚀 The feature, motivation and pitch

Motivation:
We have made great strides in enforcing code quality and consistency within our Python codebase, particularly with the recent MyPy integration. To maintain a high standard of code quality throughout the project, we should apply similar formatting and style guidelines to our C++ and CUDA code.

Proposal:
Integrate clang-format into our format.sh code quality workflow to automatically enforce consistent formatting for C++ and CUDA files. This will ensure that all code adheres to the Google C++ style guide, making the codebase more readable and maintainable.

Benefits:

• Consistent formatting across the entire codebase, including C++ and CUDA files.
• Improved code readability and maintainability.
• Reduced cognitive overhead for developers working with multiple languages within the project.

Alternatives

No response

Additional context

No response

[simon-mo]

Yes! I believe @LiuXiaoxuanPKU also mentioned something like this before. The only concern/corner case this needs to address is some "vendor" files need to skip formatting like exllama or awq kernels. (to minimize the diff)

[mgoin]

Resolved by #4722