Fix cudastack for JetPack 5.

[TurbineJoshua]

My JetPack 5 has CUDNN 8.6.0 installed.

dpkg -l | grep cudnn
ii  libcudnn8                                                   8.6.0.166-1+cuda11.4                 arm64        cuDNN runtime libraries
ii  libcudnn8-dev                                               8.6.0.166-1+cuda11.4                 arm64        cuDNN development libraries and headers
ii  libcudnn8-samples                                           8.6.0.166-1+cuda11.4                 arm64        cuDNN samples

I have confirmed that using an image built from these changes, I was able to compile PyTorch 2.3.1 from source with CUDNN support enabled:

>>> import torch
>>> torch.version
<module 'torch.version' from '/installation/.venv/lib/python3.10/site-packages/torch/version.py'>
>>> torch.__version__
'2.3.1'
>>> torch.cuda.is_available()
True
>>> torch.backends.cudnn.is_available()
True
>>> torch.backends.cudnn.version()
8600