I've come across an excellent solution for exactly this scenario.
It uses automatic gain control on the presenters' microphones, and switches the camera to whichever one is loudest.
This would probably be easiest to achieve by switching a video feed to a single capture device.
It was a few years ago, and I heard it on one of the element opie podcasts, possibly "The art of podcasting".
If I recall anything more specific, I'll post back up.
I'd love to see some more discussion here about hardware setup for producing quality recordings.