Fix --data from_HUB.zip

[glenn-jocher]

@kalenmike

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Enhanced file validation and streamlined model weight checks in YOLOv5 training script.

📊 Key Changes

• Replaced check_yaml with check_file for more generalized file validation.
• Moved the model weight file format check (.pt) to an earlier stage in the training process.

🎯 Purpose & Impact

• Purpose: To ensure user-provided files (data, config, weights) are properly validated, promoting better error handling and avoiding potential issues during model training.
• Impact: Users will experience more robust checks before training begins, potentially reducing errors due to incorrect file formats or paths. The changes will lead to a smoother training process for developers and practitioners using the YOLOv5 framework.