Text this: Bayesian causal forests with covariate balancing propensity score: a novel approach for heterogeneous treatment effects estimation