Excellent question indeed. Here we go: The stereo mix and the drum buss mix are actually -- and need to be -- discrete..that is, unrelated regarding level balancing. First you build your basic stereo drum mix. Then you send the individual elements to the drum buss pre-fader. This allows you to adjust the level to the drum buss if something is too loud or quiet after you apply the drum buss processing, so that's cool and gives you a lot of creative control.
We need the sends to be pre-fader so that a fader change to the stereo drum mix does not disturb the delicate balances achieved in the drum buss mix. This way everything is discrete and intentional.
The drum buss reverbs output directly to the drum buss. The sends to the drum buss reverb are also pre-fader for the exact same reason.
It is this combination of the dynamics of the stereo drum mix and the density and detail of the drum buss mix that makes the magic happen.
does this makes sense?