ep_return may be a numpy array in multi agent scenarios

[liuzhaoze]

Referring to here, ep_return may be a numpy array where the number of elements is equal to the number of agents, representing the return of each agent.

[MischaPanch]

Ah, I see, thanks for the hint! I haven't thought about the multi-agent case when I wrote this todo. Gonna look into it, generally the multi-agent logic and the single-agent logic shouldn't be covered by the same interfaces. It is confusing, and multi-agent remains a very niche case that should be treated separately