Describe the workflow you want to enable

I'd like to use HistGradientBoostingClassifier and HistGradientBoostingRegressor with the possibility to set interaction constraints for certain features. As said in microsoft/LightGBM#2884 (comment), it is one way to make those black boxes more intuitive and interpretable. In addition, it makes it much more easy to marginalize over those features.

Additional context

LightGBM has interaction_constraints, see their docs. XGBoost has them, see their docs.
Have also a look at the XGBoost tutorial on interaction constraints for a nice visualization and for potential benefits:

• Better predictive performance from focusing on interactions that work – whether through domain specific knowledge or algorithms that rank interactions
• Less noise in predictions; better generalization
• More control to the user on what the model can fit. For example, the user may want to exclude some interactions even if they perform well due to regulatory constraints