fix push data into buffer

com.unity.ml-agents/Runtime/Policies/TrainingPolicy.cs

                 m_buffer.Push(m_LastInfo, m_LastObservations, m_CurrentObservations);
             }
 
+            m_LastInfo = info;
+            m_LastObservations = m_CurrentObservations;
+
+                m_buffer.Push(info, m_CurrentObservations, m_CurrentObservations); // dummy next_state
-            }
-            else
-            {
-                m_LastInfo = info;
-                m_LastObservations = m_CurrentObservations;
             }
         }