Create a riichi mahjong AI to play on tenhou.net

Number of effective samples: 75110

Evaluation value: 0.8220610355708976

(1). HS

 MSE =  0.46887228095040395

(2). HS_WFW:

 MSE = 0.3700624104298077

Finish Monte Carlo simulation.

(1). Opponent model (mahjong.ai.opponent_model.py) is now under development. Note that we have already switched to opponent model in mahjong.player.py (line 262-266), which means we have intentionally turn off the original AI of the author.

(2). An example to use in reproduce_test: -m ..\DATA\riichi_mahjong\raw_data\2016120921gm-00a9-0000-ec827574.mjlog -d