The influence of rarity and expertise on pseudodiagnosticity