Typo of actor_critic.py? about minimalrl

[seungeunrho]

Oh my god... You are absolutely correct. It was a critical typo.
I revised the error and pushed it!

The reason why the original code works well is due to the small value of "n_rollout".
I originally set it as 5.
During 5 ticks, the agent state maybe does not alter enough to make the value estimate incorrect.

If you modify the n_rollout to 20 and run both versions of the codes,
only the one you suggested works properly.

Thank you again.