Enable Apple Silicon GPU Acceleration for d2l

[CharryLee0426]

Description of changes: In this Pull Request: The Apple Silicon GPU support is added to d2l. So after this commit, d2l can detect Apple Silicon directly and use Apple Silicon GPU to accelerate parallel computing. It applies to two most popular deep learning frameworks: PyTorch and TensorFlow.

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[CharryLee0426]

If anyone interested in this pull request, look at issue#2588