[QDP] GPU TF tensor support with dlpack and device placement #729
Description
Summary
This issue asks GPU tensor support using dlpack protocol (TF 2.5+) and handles TensorFlow's device placement system.
- extend encode_tf_tensor() to handle GPU tensors
- add DLPack extraction (TF 2.5+)
- handle TensorFlow device placement
- Handle TensorFlow-specific scenarios
- add stream synchronization
- some tests
If anything important is missing, please leave a comment.