Issue with the wrappers of NormalizeObservation for the env #75
Description
Hello! I am a learner studying the safe reinforcement learning. I am puzzled as to why the wrapper of NormalizeObservation in the repository uses the one of gymnasium, rather than the one of safety-gymnasium. I think the wrapper of NormalizeObservation in gymnasium does not normalize the "final_obs". It will leads to an inaccurate bootstrapped value prediction.