Model enhancement

[AyushExel]

Adds:

• model.fuse()
• model.val()
• model.visualize_preds()
• model.predict()
• improves resume
  TODO:

• Add pytest-like model tests
• Write docs

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Enhancements to the YOLO model testing and loading capabilities.

📊 Key Changes

• Added new test cases for various model functionalities.
• Expanded the YOLO class with methods info, fuse, predict, val, resume.
• Introduced model-task associations enabling clearer operations based on model types.
• Enhanced flexibility in handling model files and inputs.
• Added smart inference using the @smart_inference_mode decorator.

🎯 Purpose & Impact

• 🎨 Streamlines the testing process by modularizing tests and adding new ones for major functionalities.
• 🚀 Enables a more robust usage of the YOLO model class with improved methods for info logging, model fusion, prediction visualization, validation, and resumption of training.
• 🌐 Fosters ease of integration with diverse model formats and devices for wider applicability and superior performance.
• 🧑‍💻 Provides developers with more intuitive control over tasks and models while enhancing the usability of the model for end users across various applications.

[AyushExel]

@Laughing-q I've added a model tester file that tests all functionality of the model with one script. Once we have the assets in place we can then start using it in the ci.

I'll merge it like this wihtout adding preds post-process as I also cannot see how users will use it. I think .predict is what we want users to use