A easy to approach demo to do with ML

• Shows how to train a good model
• Link to the benchmark paper
• Multi-agent since that is the point of SMARTS
• Take current PR Multi agent reinforcement training demo #797 and clean up for inclusion before adding HPC
• Add HPC