Add colorstr()

[glenn-jocher]

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Enhanced anchor scaling and improved command-line feedback in the YOLOv5 training script.

📊 Key Changes

• Anchor scaling logic in train.py is modified to factor in the number of output layers.
• Autoanchor analysis in utils/autoanchor.py now includes more descriptive prefix output for better readability.
• The git status check in utils/general.py now checks for the existence of .git and Docker environments more robustly.
• Loss calculation in utils/loss.py simplifies scaling factors for lbox and lcls losses.
• utils/torch_utils.py includes a new method to describe the git state alongside selecting a computing device.

🎯 Purpose & Impact

• 🛠️ Improves anchor scaling to potentially boost object detection performance.
• 📢 Enhances user feedback during anchor analysis with colored print statements.
• 🔍 Provides better Git repository status checks to inform users when updates are available.
• 🧮 Optimization of the loss computation could lead to more effective training.
• 📊 Git descriptors in device selection output offer users clearer version tracking information.