Okay, I didn't mean to scare you. I know nothing about how controllers and throttles work, but I think there is a +5V and a (common) ground from the controller to the throttle. If these are tied together, but the controllers get their power from separate batteries this would result in both batteries being grounded together but the +5V circuit is probably isolated (via diodes?) from the main battery positive. So my guess is no real problem. Maybe one of the experts will chime in on this issue again.
In any case, I still think paralleling the batteries makes sense because that keeps the voltage to each controller and motor system the same. This didn't present a problem on my initial systems, since I was using SLA at the time. However, when I used LiFePO4 batteries, I did bypass the BMS during discharge, making sure they were at the same voltage before hooking them together in parallel. A better solution would have been a single battery pack large enough to handle both motors.
I knew you said your battery packs were different sizes, but I just looked at your first post and I see they are 15AH and 16AH Pings, which are almost the same. Since LiFePO4 tends to operate at a pretty constant voltage until near the end, you probably won't have any problems unless you try to run them clear out, which isn't a good thing to do anyway. Your meter may tell the story, because if the voltage of one battery drops below the other the motor with the higher voltage will try to run faster, using more juice, which would then balance the two batteries. i.e., they may stay at the same voltage because of the mechanical link with no need of an electrical (parallel) link.
So in conclusion, sorry about being so long winded, what you are doing right now may make perfect sense and work better than any other technique. Let us know.
EDIT: What Fechter wrote while I was writing this is what I have seen before, i.e. not paralleling the 5V throttle wires together and instead just using the 5V wire from one of the controllers. I have never tried doing that, but if it could potentially save a controller may be well worth doing. I no longer use any multi-motor setups, so will not be doing any such tests myself.