Add py-torch to container so that libtorch binaries are accessible for running inference
No changes between version 8 and version 8
