Limit tensorflow version and add checks

[Corallo]

Exporting to tflite format fails when the activation function is a leakyReLU / ReLU.

The bug was reported here: #12489
The problem seems to be with the tensorflow version, that has to be <=2.13.1
The other repository (https://github.com/ultralytics/ultralytics) have had a similar patch ultralytics/ultralytics#6466
I am applying the same here to constraint tensorflow version.

copilot:all

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Adjustment to TensorFlow compatibility in YOLOv5 export functionality.

📊 Key Changes

• Added a version check in export.py to advise on a known issue for TensorFlow versions greater than 2.13.1.
• Updated the requirements.txt file to restrict TensorFlow version to 2.4.0 or above and up to 2.13.1.

🎯 Purpose & Impact

• Ensures users are aware of potential problems when exporting models with TensorFlow versions newer than 2.13.1. 🛠️
• Prevents installation of incompatible TensorFlow versions that could cause export issues, enhancing stability and user experience. 🧰
• Users running yolov5 exports with TensorFlow will have a smoother experience with controlled version compatibility. ✅

[github-actions]

👋 Hello @Corallo, thank you for submitting a YOLOv5 🚀 PR! To allow your work to be integrated as seamlessly as possible, we advise you to:

• ✅ Verify your PR is up-to-date with ultralytics/yolov5 master branch. If your PR is behind you can update your code by clicking the 'Update branch' button or by running git pull and git merge master locally.
• ✅ Verify all YOLOv5 Continuous Integration (CI) checks are passing.
• ✅ Reduce changes to the absolute minimum required for your bug fix or feature addition. "It is not daily increase but daily decrease, hack away the unessential. The closer to the source, the less wastage there is." — Bruce Lee

[glenn-jocher]

@Corallo hi there! Thanks for reporting this issue. We appreciate your diligence in tracking down the root cause and providing a solution. You can limit the TensorFlow version by adding tensorflow<=2.13.1 to your requirements.txt file and perform a version check in your code using tf.__version__ to ensure compatibility. Feel free to visit our Ultralytics Docs for further guidance. Your contribution is valuable to the YOLO community! 🚀

[Corallo]

@glenn-jocher The issue doesn't apply only to my customization, but also to the version you have in ./models/hub/yolov5s-LeakyReLU.yaml
Exporting this version to tflite --int8, as mentioned in the issue, results in a broken network.
This PR fixed the issue with that one too.

[glenn-jocher]

@Corallo Appreciate your insights! It's great that the fix applies to both customizations and the provided YOLOv5s-LeakyReLU model. Maintaining compatibility across different configurations is crucial. We'll look into updating the default configuration. Thank you for your continued contributions! 👍

[glenn-jocher]

@Corallo PR merged! Thank you for your contributions.