get_nowait in env_manager

ml-agents/mlagents/trainers/env_manager.py

                 # We make sure to empty the policy queue before continuing to produce steps.
                 # This halts the trainers until the policy queue is empty.
                 while True:
-                    _policy = self.agent_managers[brain_name].policy_queue.get(
-                        block=False
-                    )
+                    _policy = self.agent_managers[brain_name].policy_queue.get_nowait()
             except AgentManagerQueue.Empty:
                 if _policy is not None:
                     self.set_policy(brain_name, _policy)