Use reward sum

ml-agents/mlagents/trainers/trajectory.py

             )
             agent_buffer_trajectory["team_rewards"].append(teammate_rewards)
             team_reward = teammate_rewards + [exp.reward]
-            agent_buffer_trajectory["average_team_reward"].append(
-                sum(team_reward) / len(team_reward)
-            )
+            agent_buffer_trajectory["average_team_reward"].append(sum(team_reward))
 
             # Next actions
             teammate_cont_next_actions = []