Improved non-latin Annotator() plotting

[glenn-jocher]

May resolve #7460

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Improved support for non-Latin text in training data labels and UI elements.

📊 Key Changes

• Addition of a check for non-ASCII class names in data labels, prompting the download of a compatible font if necessary.
• Introduction of a progress parameter to the check_font function to show download progress.
• Modification to Annotator class to better handle non-Latin text by selecting the appropriate font.

🎯 Purpose & Impact

• 🚀 Purpose: The primary goal is to ensure that YOLOv5 can effectively work with datasets that have non-Latin scripts (such as Asian, Arabic, Cyrillic, etc.) by automatically checking and downloading a font that supports these scripts, thus avoiding errors in label visualization.
• 💡 Impact: Users working with international datasets will experience smoother setup and training processes, as well as improved display of non-Latin characters during training, evaluation, and inference stages. This will enhance the usability of YOLOv5 for a global user base.