In the presence of conflicts, the phone could prompt the user for audio source, then persist it in software until the next time they have a new conflict. It really seems to be an issue when you go from no streaming audio to streaming audio, from the UX perspective, so that should minimize the number of times that prompt appears. Could also solve it by the same way we solve SSID preferences: a static list of paired devices with priority orders and an accessible menu to select a specific one, and then persist that until you change your streaming state again.