Update DetectMultiBackend for tuple outputs

[glenn-jocher]

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Refinement of model output handling for better compatibility across different platforms.

📊 Key Changes

• Consolidated the condition that checks and unpacks the tuple output from the model.
• Removed individual tuple unpacking conditions in each platform-specific code block (PyTorch, TorchScript, ONNX Runtime, etc.).

🎯 Purpose & Impact

• Consistency: Streamlines output handling by ensuring all platform-specific code paths adhere to a unified approach for unpacking tuples.
• Code Simplicity: Simplifies readability and maintenance since there's now only one instance where the tuple is unpacked, reducing redundancy.
• User Benefit: Users can expect consistent behavior across different deployment platforms, potentially reducing platform-specific bugs.