Refactor code

[glenn-jocher]

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Enhancements and code tidying for better performance and readability.

📊 Key Changes

• Updated documentation in benchmark/README.md to improve clarity on matrix multiplication and parallelization.
• Removed unused imports like pytest and jinja2.StrictUndefined to clean up the code.
• Simplified mathematical operations across various test files for clarity and efficiency.
• Streamlined the calculation process in some functions by using straightforward operations.
• Deleted onnx_profile.py, removing support for ONNX profile calculations, focusing the project on its core functionalities.
• Incremented the version number from 0.2.4 to 0.2.5, indicating new changes or improvements.
• Improved code quality by removing unnecessary comments and optimizing the use of control structures like for-loops.

🎯 Purpose & Impact

• For Users: These changes aim to make the tool more efficient and easier to understand, potentially speeding up calculations and reducing errors in profiling neural networks.
• For Developers: Codebase cleanups and removals of deprecated or unused features make it easier for future contributions and maintenance.
• Removing ONNX profiling support may focus the project more directly on its core offerings, although it might impact users relying on this feature.