I also assume the cradle does not have an address, but instead is just looking for specific messages on the bus. If that's the case, two (or more) cradles would work fine, but the same wheel action would be directed to all the cradles at the same time, which would probably not be what the user wants
Pretty sure that's how it works, using the model of the CANbus, which I'm familiar with. Both units would respond to a control command with unpredictable results. For example, a "Zoom +" command can mean different things from zoom the map to answer the phone, depending on the context. Sounds like it would be very confusing.
(assuming the cradle doesn't eat the message, in which case the wheel would control only one cradle, most likely the one on the bus closest to the wheel wiring).
Speed of electrons in copper being what it is, I don't think one device could detect and "eat" a bus broadcast before another one records the event. The complicated part that might cause some concern, however, is broadcasts from the Nav(s) back to the bike, e.g. the current time to set the bike's clock. The bike is probably just looking for a given packet header and there would just be twice as many as usual so, thinking about, maybe less of a problem.
If you wanted to run two units, taking one off of the bus would be the best bet. Both cradles could still be fully functional LINbus-connected but you could just apply a small dot of electrical tape to the LINbus contact (a single pad) on the back of the Nav you didn't want to respond to remote commands. You might want to insulate the audio-out pad as well.
My head hurts. I'm thinking I'll stick with one Nav.