Add ConfusionMatrix normalize=True flag

[glenn-jocher]

Per #3533

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Improved customizable confusion matrix plotting in YOLOv5.

📊 Key Changes

• Added normalize parameter to the plot method within utils/metrics.py.
• Enabled optional normalization of the confusion matrix before plotting.

🎯 Purpose & Impact

• The normalize parameter allows users to choose whether or not they want to see their confusion matrix normalized, which provides flexibility in analyzing model performance.
• This can be especially useful for users wanting to understand the raw number of cases for each prediction category, or compare different models' performance more directly.
• The impact is mainly on improving the usability of the model evaluation tools, potentially making it easier for developers and data scientists to tune their models and communicate results. 📈