Update requirements.txt with tf-cpu and tf-aarch64

[glenn-jocher]

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Enhanced TensorFlow compatibility for YOLOv5 TFLite export

📊 Key Changes

• Updated requirements.txt to clarify TensorFlow requirements for TFLite export.

🎯 Purpose & Impact

• 🎨 Simplicity for Users: Helps users understand that they can use different TensorFlow packages (tensorflow-cpu, tensorflow-aarch64) based on their hardware for TFLite model exports.
• ⚙️ Flexibility: Provides users with more options for TensorFlow installation, accommodating a variety of system architectures.
• 🚀 Accessibility: Makes it easier for users to export their YOLOv5 models to TFLite format, potentially expanding usage to more platforms, such as lightweight or mobile devices.