(especially about mutual shutdown circuitry for bipolar supplies..)
What should be done is an interesting topic.
A lot of professional Lab quality power supplies have a tracking mode which typically is used when you want a bipolar +/-V supply. If you haven't used one of these before the voltage setting on the master supply controls the outputs of both slave and master DC outputs. The pot on the slave channel is disabled. In non-tracking mode both channels behave independently and identically and the master/slave labeling means nothing.
In one configuration the second channel (slave) tracks the first (master) by literally tracking the voltage on the output terminals of the master. When you overload or short the master the voltage drops and the slave will follow. However when you overload or short the slave, the slave voltage drops but the master does not follow. The behaviour is asymmetrical.
In a second configuration the second channel (slave) tracks the first (master) by following the voltage *pot* setting. When you short or overload either channel only the voltage on that channel is affected.
I reviewed a number of professional power supplies and both behaviours are out there. As far as I can rememeber there's none which pull both rails down if *either* one of the rails drops, like a mutually tracking idea. No doubt that's because of the messy analog electronics required to handle this off the radar case. (There was one Tektronix power supply which had a lot of funky stuff going on but I can't rememeber if it did mutual tracking at the end of the day.)
All being said you might not want or care if the supplies to mutually track!
Anyway the take home message is there's no standard for this behaviour so there's no point enforcing such requirements on a simple supply because not even the big-boys are consistent.
Having a tracking mode can be convenient.