Fix val.py Ensemble()

[glenn-jocher]

May resolve #7475

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

This PR enhances model loading and validation checks for the YOLOv5 object detection suite.

📊 Key Changes

• Improved attribute assignment consistency when creating an ensemble of models.
• Added an assertion to verify that all models in an ensemble have the same number of classes.
• Updated validation code to use the correct attribute for the number of classes when checking model-data compatibility.

🎯 Purpose & Impact

• 🧩 These changes ensure more robust and error-proof model ensembling by standardizing key attributes across all models in the ensemble.
• 🚦 The new assertion prevents potential errors due to class count mismatches in ensemble models, enhancing the reliability of the system.
• 🕵️‍♂️ By fixing the validation check in val.py, users are less likely to encounter issues when their model’s class count does not match the dataset’s class count, improving user experience and model accuracy.

[glenn-jocher]

Verified working: