Is it possible to evaluate multiple cost functions at once, then produce the next set of likely parameters to evaluate, a.k.a. batch updating the internal model state with parallel function evaluation?
No, that's not supported in BADS. There is plenty of work on batch/parallel evaluation for Bayesian optimization, but it wouldn't be straightforward to implement as of now.
Is it possible to evaluate multiple cost functions at once, then produce the next set of likely parameters to evaluate, a.k.a. batch updating the internal model state with parallel function evaluation?