First CRAN release.
pareto_design(): finite alternatives, minimised objectives with known
bounds, epsilon-dominance, alpha shared by a union bound over the K x m
confidence sequences, boundary from seqbench, cost and budget.initialize_pareto(), update_objectives(): one anytime-valid confidence
sequence per alternative and objective, running intersections, auditable
history and stopping reason.plausible_pareto_set(): estimated frontier, plausible Pareto set, certified
optimal and certified dominated alternatives (each with the alternative that
dominates it); epsilon-ties keep one representative.choose_next_evaluation() and pareto_run(): adaptive choice of the next
alternative by confidence-box width per unit cost, and a run loop with a
user simulator.pareto_report(); S3 class paretoinfer_state with print, summary,
tidy, glance, autoplot / plot_uncertain_frontier().