Pairwise Causal Discovery Algorithms in pgmpy

This project adds pairwise causal discovery methods to pgmpy, focusing on inferring causal direction between two variables from observational data. It includes implementations of ANM, IGCI, and Bivariate LiNGAM under a simple, unified interface, along with an HSIC-based independence test and basic benchmarking using the Tübingen CauseEffectPairs.