Investigate possibility of installing nvidia runtime via standard install pipeline

[johnhbenetech]

Currently a user who wants to use their GPU must remember to run install_nvidia_docker.sh before running the tool in order to access the runtime.

We should investigate the potential to run this script automatically during the build process if the user chooses GPU during the 'make' flow.