comment out comms in ppo optimizer

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

         if len(memories) > 0:
             memories = torch.stack(memories).unsqueeze(0)
 
-        comms = self.policy.get_comms(
-            comm_obs[0],
-            masks=act_masks,
-            memories=memories,
-            seq_len=self.policy.sequence_length,
-        )
-        # this is a little bit of a hack but is whats recommended in the
-        # gumbel softmax documentation
-        one_hot_diff_comms = obs[-1] - comms[1].detach() + comms[1]
-        obs[-1] = one_hot_diff_comms
+        #comms = self.policy.get_comms(
+        #    comm_obs[0],
+        #    masks=act_masks,
+        #    memories=memories,
+        #    seq_len=self.policy.sequence_length,
+        #)
+        ## this is a little bit of a hack but is whats recommended in the
+        ## gumbel softmax documentation
+        #one_hot_diff_comms = obs[-1] - comms[1].detach() + comms[1]
+        #obs[-1] = one_hot_diff_comms
 
         log_probs, entropy, values = self.policy.evaluate_actions(
             obs,