Improved dataset error introspection

[glenn-jocher]

Intended to help #8090

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Enhanced dataset validation and automatic configuration in the YOLOv5 framework.

📊 Key Changes

• Combined dataset downloading, checking, and unzipping into a single function for simplicity.
• Added checks within the check_dataset function to ensure provided data has necessary keys like 'train', 'val', 'nc' (number of classes), and assigns default 'names' if missing.
• Removed redundant assertions and name assignments outside the initial data checks to streamline the code.

🎯 Purpose & Impact

• The changes improve user experience by performing more comprehensive and upfront checks on datasets, which helps prevent common errors during training.
• Default class names are now automatically generated if not provided, simplifying dataset setup for new users.
• These improvements result in a more robust and user-friendly setup process when preparing data for model training, enhancing overall accessibility for both new and experienced users of YOLOv5. 🚀