If it doesn't rotate, a castle nut is not needed. In this case, the bolt is under no rotational force, or at least nothing significant. It clamps on the bearing race.
I've been frustrated many times by people that have A&P tickets that don't understand where "designed" bearing surfaces/points are in this type of assembly. Some even worked for me for a while who insisted on leaving a bolt a little loose with nylock nut installed. The replacement then of unnecessarily worn components can get real expensive, particularly on helicopters.
The pic above is of a rod end that contains a bearing that only works if that ball is truly captive forcing the designed bearing surface to in fact be the bearing surface. Nothing wrong with a torque stripe on the nut to bolt threaded end if desired.