Actually the mixing left and right into center left/right probably wouldn't work out like I'd want it. What I would like is better dialog steering, but in most soundtracks I'm pretty sure they mostly dump any dialog near the front into the center.
The metadata deal is IMO probably as close to the 'final' word as we'll get in the foreseeable future. It's not an original idea, as you already know. I think I first saw it mentioned in relation to a PC gaming API IIRC. In any case, locating the source of sound in at least 3 degrees of freedom would allow the theater processor to do as good a job as possible with whatever the input is. For instance if you have a 240.2 system with a full hemisphere of transducers and I have a speaker in each corner we could both get the most out of our systems.