ULearningAgentsTrainingEnvironment::GatherAgentRewards

This callback can be overridden by the Trainer and gathers all the reward values for the given set of agents.