This is the only rational reason for redundant busses: when you need separate outputs. Examples include headphone mixes, using room EQ for monitoring, and routing to alternate speakers.
 
Regarding the -0.1 dB ceiling, that's only justified if you are mastering in place for CD distribution, the presumption being that listeners will be using high-quality playback systems that can easily handle any unexpected analog peaks. If you are mastering for lossy formats such as MP3, those "intersample" peaks are far more likely to occur, and listeners are far more likely to be listening on battery-powered devices that can't handle them. And of course, if you plan to have a professional master your mix, he'll likely reject anything that comes in limited to -0.1, or even -1 dB.

bitflipper
This is the only rational reason for redundant busses: when you need separate outputs. Examples include headphone mixes, using room EQ for monitoring, and routing to alternate speakers.

"Redundant" yes.
But functional extra busses before the final "A" are quite rational and functional for others.

I started using a submaster or 2 bus a couple of years ago.  Sometimes I'll run some fx on the 2 bus that has most instruments, or all instruments, or everything.  I compress that bus to help glue everything going there.  I may even have the instruments on the 2 bus and use the vocals or bass going to the master as a sidechain against the 2 bus.  It just adds another layer of possibilities if I feel like using it.  I master as a separate process, even tho the mixes are mostly done for each project.  I usually have the master bus just manage limiting, dithering and output monitoring.