Best way to understand this is to go back with me to our sampling session when we recorded to room's acoustics for the IR in Altiverb.
mono to stereo
We put a single speaker in the center of the stage and two mics in the audience.
This results in a mono to stereo IR.
When you playback your audio through this IR in Altiverb, it will sound like your audio comes out of that single speaker from the center of the stage (mono!) picked up with two microphones (stereo!).
This IR gives a stereo reverb, however the input is (mixed to) mono, as there is only a single source on stage.
When you want to process a vocal (mono) or direct recorded electrical guitar or something where the panning is not important the mono to stereo IR will do just fine (pro: costs half of the processing then a stereo-to-stereo IR, so it is more efficient in cpu and memory).
stereo to stereo
back to recording the room
this time we put two speakers on stage, one on the left and one on the right side of the stage and we record again with the two mics in the audience.
When you playback your stereo audio through this IR in Altiverb your stereo signal comes out of these two speakers.
Easy to picture here that your original panning will be maintained using a stereo to stereo IR.
So basically it comes down to this:
is your input stereo or is it panned in stereo ?
use a stereo - stereo IR, else you will loose the stereo information in the reverb (although the reverb is stereo).
Sometimes even when using stereo panned sources a mono to stereo IR can be preferred as you do not want the panning to be reflected in the reverb levels. But this is a mixing decision which depends on taste mostly (and music style) and I dare not to advice on that :-)