Depends on the specific setup.
You may be able to set up a simulation on http://ebikes.ca/tools/simulator.html to see what differences there are in each method, using the system B in add mode, since they have independent throttles. I did a quick test of it and it seems like there's a problem with using just one system with 0% throttle on the other--seems to require at least 3% throttle on the "unused" system to get a result, but I might not have set it up right.
as far as "bad" for the non running system...if they are not the same motor and/or wheel size, such that the max speed of one is much higher than the other, then there could be issues when above the max speed of the slower one, as it may begin adding drag to the bike, depending on the system setup. If the difference is large enough, it potentially could generate high enough voltages just from spinning to damage it's controller if the battery has an HVC that turns it's output port off and the voltage without load is high enough to be greater than the controller can handle. Same can happen even with load if it's high enough.
If the systems are the same on each wheel, then it shouldn't cause problems to have one on and one off or unused; efficiency/range might be another matter you'd need to test in simulation or reality.
There's been threads about this general topic a few times over the years, that will have more information in them, if you poke around under 2WD, dual motor, etc.
