Why the reward function must be called before updating the observation in PointWiseEnv.py?

![1659071920738](https://user-images.githubusercontent.com/57398559/181687904-9898a57a-474b-423f-aec2-0e41db1f6709.png)

Shouldn't the reward be an evaluation of the observed new state S(t+1)?
For example，[openai_gym_cartpole](https://github.com/openai/gym/blob/master/gym/envs/classic_control/cartpole.py)