rejection sample dynesty posterior

In bilby!1203 (merged) we switched to rejection sampling the nested samples to get posterior samples. It would be nice to also have this option in pBilby.