added ppo/sac_policy attributes to keep up with master

ml-agents/mlagents/trainers/ppo/trainer.py

         for _reward_signal in policy.reward_signals.keys():
             self.collected_rewards[_reward_signal] = {}
 
+        self.ppo_policy = policy
+
         return policy
 
 

ml-agents/mlagents/trainers/sac/trainer.py

                     self.update_buffer.num_experiences
                 )
             )
+
+        self.sac_policy = policy
+
         return policy
 
     def update_sac_policy(self) -> None: