Potential performance issue: GroupBy.quantile slow in pandas below 2.1

[TendouArisu]

Issue Description:

Hello.
I have discovered a performance degradation in the GroupBy.quantile function of pandas version below 2.1. And I notice in src/requirements.txt, shows that it depends on pandas version 2.0.0. I am not sure whether this performance problem in pandas will affect this repository. I found some discussions on GitHub related to this issue, including #51385 and #51722.
I also found that src/preprocess/data stats.ipynb used the influenced api. There may be more files used the influenced api.

Suggestion

I would recommend considering an upgrade to a different version of pandas >= 2.1.
Any other workarounds or solutions would be greatly appreciated.
Thank you!