Fix academy reset out of order

python/ppo.py

 Options:
   --help                     Show this message.
   --batch-size=<n>           How many experiences per gradient descent update step [default: 64].
-  --beta=<n>                 Strength of entropy regularization [default: 5e-3].
+  --beta=<n>                 Strength of entropy regularization [default: 2e-3].
   --buffer-size=<n>          How large the experience buffer should be before gradient descent [default: 2048].
   --curriculum=<file>        Curriculum json file for environment [default: None].
   --epsilon=<n>              Acceptable threshold around ratio of old and new policy probabilities [default: 0.2].

unity-environment/Assets/ML-Agents/Examples/GridWorld/Scripts/GridAgent.cs

     // to be implemented by the developer
     public override void AgentStep(float[] act)
     {
-
         reward = -0.01f;
         int action = Mathf.FloorToInt(act[0]);
 

unity-environment/Assets/ML-Agents/Scripts/Academy.cs

         currentStep = 0;
         episodeCount++;
         done = false;
+        AcademyReset();
+
 
         foreach (Brain brain in brains)
         {
 
-        AcademyReset();
     }
 
     // Instructs all brains to collect states from their agents.