CupGame and CleanupTask not returned in `env.step`

[jaydenmilne]

Describe the new feature you'd like env.step() returns a 4tuple of (state, reward, terminal, env_info). Until now env_info has just been None, but with the CupGame and the CleanupTask we should probably be returning information about the worlds in those tasks.

• [ ] Verify this is how gym uses the 4th parameter
• [ ] In cupgame or trash scenarios, the 4th parameter returns the current state
• [ ] Decide if this is even worth doing

This is fairly low priority