Support for min_data_in_leaf

Could a feature be added to support requiring a minimum sample size per leaf? This feature is convenient in regulated models, where terminal nodes should not be represented by less than N observations. Lightgbm ([min_data_in_leaf](https://lightgbm.readthedocs.io/en/latest/Parameters-Tuning.html)), catboost([min_data_in_leaf](https://catboost.ai/en/docs/references/training-parameters/common#min_data_in_leaf)), and sklearn([min_samples_leaf](https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.GradientBoostingClassifier.html)) support this behavior. The request has been described previously on [stackoverflow](https://stackoverflow.com/questions/69002149/can-we-set-minimum-samples-per-leaf-in-xgboost-like-in-other-gbm-algos). Thanks in advance for a redirect if there is already a technical discussion elsewhere.