Thanks Ash. Very helpful. I had not realized that ReaVerb sums to mono before splitting the signal to stereo. I'll try the dual reverb trick. Would this be the procedure? 1. Set up two reverb tracks, panned left and right. 2. Set up two separate sends from a mono track, one to each reverb track. 3. Adjust the panning position of the reverb-to-master return by balancing the send levels from the mono tracks to the reverb tracks. Am I on the right...er...track...here?
If you stop and think for a moment about how the impulse is recorded, I think you'll see that it kind of can't be any other way. Say you set up a stereo pair of microphones and pop a balloon. The mics don't have any idea where the balloon is in relation to them, just what actually hits their diaphrams. Pop two balloons at once. The mics still don't know what part of what they're hearing come from the one on the left or the one on the right. No way to sort it out.
So instead you have to pop the balloon on the left, capture the response in both mics, then repeat for the balloon on the right. Then when you use the IR, you have to send the things you want to sound like they are coming from where the left balloon was to through the one IR, and other things through the other. Or, if the room is relatively symmetrical or you're just not that worried about it, you use the left one for both, but flip the output channels of the one.
This, of course, doesn't help you realistically simulate positions
in between where those two balloons were. You end up with an effect very much like if you mixed the thing (based on your the levels and panning of the sends) into a pair of speakers in the room, with all the "phantom center", not-exactly-realistic positioning information that entails. Frankly, this is going to be true with any true stereo reverb.
There are a very few real ray-tracing room simulators out there where you can actually place a particular source in a particular spot in respect to the "microphones". These require, of course, a separate input for each individual source, require a whole lot of resources, and are generally pretty expensive, though I think either one of the last Cakewalk ProAudio or first Sonar versions had one that didn't work great. It could be done with IRs, but would be tedious, and basically require an instance of ReaVerb for every track, and... I have been tempted to try it with Voxengo's Space Designer, but it's really probably more work than it's worth. Since I started recording, I've had the vision of pumping each track through an actual speaker in an actual room and mixing by moving them around. That would probably be easier.
Anyway. You can put ReaVerb on two separate tracks, and use two separate sends, but that gets kind of wonky fast unless you're using post-fader sends rather than messing much with panning on the sends themselves. The procedure for setting it up on one track is actually in the Reaper User Guide as an example of how to use the plugin pins and multi-channel routing and is probably worth working through just for that.