The three of them have to be used as a tool of analysis.
Balacing only around rDPS means that if the top jobs are the "selfish" ones such as SAM or BLM, then they are good for every piece of content in any comp.
For these jobs, rDPS = nDPS. Their rDPS will be the exact same (ignoring variance) whether their group has no buffs or optimized buffs. They're also not punished in this metric if they don't align their cooldowns within buffs as only aDPS tracks that although partially (single target buffs have to be manually checked through rDPS).
Therefore, if only rDPS is considered for balance, then these jobs are the best in every single scenario on top of getting the highest benefits out of buffs.
aDPS also depends on composition, so while it is a relevant metric to look at, it can vary a lot in a similar way that rDPS does for buffing jobs.
The metric that will tell you the individual performance of a person without taking other people from the party into account is nDPS.
Jobs that are strong in the nDPS/aDPS department should be lower in the rDPS department and vice versa, because the higher the percentile, the more damage players will squeeze out of buffs and so the more they will feed into others' rDPS. For example, since SAM is the job that benefits the most from buffs, it should be higher in aDPS than BLM, but lower in nDPS since the latter does not benefit as much from them (though it still does).
Balance seems to be made around the average player, not the very top (95+ or more).
You can see this when checking more accessible fights (i. e. those cleared by many people of different skill) such as P5S. MCH is the top ranged/caster job in the rDPS charts in the overall percentile and until the 80th because it doesn't rely on others for its damage and actually overtakes some melee jobs (NIN and DRG, so the ones with the lowest personal damage) in both aDPS and nDPS until higher percentiles. In fact, MCH's nDPS is still higher than DRG's in the 90th percentile.