Bayesian Regression for Combinatorial Response Data // Bayesian regression whose response data are integer-valued vectors subject to combinatorial constraints in the form of Ay=b. Implements the Metropolis-Hastings-within-Gibbs sampler of Zheng et al. (2026+) doi:10.48550/arXiv.2504.11630, an unconstrained probit baseline for comparison, benchmarking helpers, MCMC and regression diagnostics (effective sample sizes, split-Rhat, structured reports), plotting methods (trace, autocorrelation, violin, effective-sample-size, sampling-efficiency, residual heat map), and utilities for constraint validation (total unimodularity, feasibility) and data simulation.