[Documentation] onnxruntime 1.13.1. Proper cuda version

### Describe the documentation issue

Hello,

Gents, could you please advise?
What is the proper cuda version for building and running onnx 1.13.1?

I always looked as a reference to the following [link](https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html#requirements).
Here at the 'CUDA' column it states CUDA version 11.6 and the release notes are also saying "Official ONNX Runtime GPU packages now require CUDA version >=11.6 instead of 11.4."
But in the 'Notes' section we still can see libs related to 11.4 and at the [Dockerfile.cuda](https://github.com/microsoft/onnxruntime/blob/v1.13.1/dockerfiles/Dockerfile.cuda) we still can see
```FROM nvcr.io/nvidia/cuda:11.4.2-cudnn8-devel-ubuntu20.04```

So the Q is: what is the proper configuration of CUDA and CUDNN for 1.13.1 to build and run?
Somethin like that?
[libcudnn8-dev_8.5.0.96-1+cuda11.7_amd64.deb]
[libcudnn8-dev_8.6.0.163-1+cuda11.8_amd64.deb]
[libcudnn8-dev_8.7.0.84-1+cuda11.8_amd64.deb]

### Page / URL

https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html#requirements

Start with the CUDA requirements section at the linked documentation page and compare it with dockerfiles/Dockerfile.cuda at the v1.13.1 tag. Reconcile the CUDA and cuDNN versions shown in the table, notes, release information, and Dockerfile, then make the supported build and runtime configuration unambiguous.

Written by the indexing model from the issue text.