Fix one more np float32 issue

ml-agents/mlagents/trainers/components/reward_signals/extrinsic/signal.py

         return RewardSignalResult(scaled_reward, unscaled_reward)
 
     def evaluate_batch(self, mini_batch: Dict[str, np.array]) -> RewardSignalResult:
-        env_rews = np.array(mini_batch["environment_rewards"])
+        env_rews = np.array(mini_batch["environment_rewards"], dtype=np.float32)
         return RewardSignalResult(self.strength * env_rews, env_rews)