Add pytorch-notebook image variants with cuda 11 and 12 (x86_64 versions only) (jupyter#2091)

* feat: build cuda variants of pytorch

* feat: build with variant tag

* style: remove unused import

* refactor: rename get_prefix params

(cherry picked from commit 12b50af)

* revert: drop ROOT_CONTAINER addition from Makefile

(cherry picked from commit f423145)

* style: use consistent three empty lines in Makefile

(cherry picked from commit 446b45a)

* refactor: add default value for parent-image

(cherry picked from commit 32955ce)

* revert: use original workflow structure

(cherry picked from commit 68c6744)

* refactor: use single build image step

(cherry picked from commit 5f1ac0a)

* fix: run merge tags regardless of repository owner

(cherry picked from commit 3fce366)

* refactor: build cuda12 instead of cuda tag

(cherry picked from commit 217144e)

* docs: add note about CUDA tags to documentation

* refactor: add default value for variant in build-test-upload

* refactor: swap ordering of cuda11/cuda12 variants

* refactor: remove optional str type in arg parser

* fix: add proper env variables to CUDA Dockerfiles

* fix: remove CUDA build for aarch64

* fix: use latest NVIDIA documentation link

* fix: skip aarch64 tags file for CUDA variants

---------

Co-authored-by: zynaa <[email protected]>

Author: 2 people

docs/using/selecting.md

@@ -205,6 +205,8 @@ It contains:
 - Everything in `jupyter/scipy-notebook` and its ancestor images
 - [pytorch](https://pytorch.org/) machine learning library
 
+> **GPU Acceleration:** Append a CUDA version prefix (`cuda11-` for CUDA 11 or `cuda12-` for CUDA 12) to the image tag to allow PyTorch operations to use compatible NVIDIA GPUs for accelerated computation.
+
 ### jupyter/datascience-notebook
 
 [Source on GitHub](https://github.com/jupyter/docker-stacks/tree/main/images/datascience-notebook) |