- refactored learning function

- remove state_last, use state instead
- added episode history

a_player.py

@@ -10,7 +10,6 @@ class APlayer(object):
 			self.other_mark = 'X'
 
 		self.state = None
-		self.state_last = None
 
 	def set_debug(self, with_debug):
 		pass
@@ -22,10 +21,10 @@ class APlayer(object):
 		state.reshape(state.size)[move] = self.mark
 		return state
 
-	def reward(self, value):
+	def new_game(self):
 		pass
 
-	def new_game(self):
+	def end_game(self, reward):
 		pass
 
 	def has_won(self, state: np.array):