Implement a basic DQN using stable baselines.

[mrcolo]

I think it would be cool to start exploring AIs with a DQN.

You can use the framework stable_baselines to implement the algorithm.

For now, you don't need to know how it thinks, but just be able to implement an interface that gives us access to a DQNagent. You can also set up a few helper functions that will log stats about the DQN.