In this phase, we will expand the number of benchmark scripts and pipelines to cover the following: - implementation of multi-node lightgbm training and inferencing - implementation of xgboost, catboost, and others - implementation of more public datasets, data processors

Milestone: an end-to-end benchmark pipeline running on 3+ public datasets, comparing 3+ tree-based frameworks.