How should we estimate diversity in the fossil record? Testing richness estimators using sampling-standardised discovery curves