You are right that they would be duplicating, and not the same way or amount. You can get away with it probably though it is theoretically incorrect to do, at least as far as I can determine.
I have come to the conclusion that it is absolutely wrong to use any dry signal. I know that may sound crazy, but you are abusing the whole theory behind altiverb if you do. Because you are no longer placing your source within the impulse, just the tail. That is stupid (though of course doing something stupid sometimes works). The entire sound should be processed by the convolution or you are crudely abusing the software as if it is some new-fangled version of a hardware reverb unit.
This includes the positioning. The idea behind the stage positioning is to allow you to avoid using dry signal for positioning. The sound is still convoluted, but it is in the place where the icons are located and has the frequency coloration of the impulse.
This is complicated by the fact that if your original sound is stereo, it already has placement embedded within it. So in that case - a stereo source - you MUST NOT use stage positioning changes, but instead use the default location - or perhaps spread apart further - but NOT adjusted from one side to another, because that is an extremely complex distortion of the audio. The most direct use of the convolution principle, in other words the most accurate, is in putting the already panned dry source through the convolution processing. If you have a mono source - which almost no one uses apparently - then you can scoot it all over the stage with no problems. Also, if you have an unpanned stereo ensemble, you could move it on the stage as well - for example with a violin ensemble that you are shrinking down in size and placing over a little to one side. But anything with positioning already in place is being redundantly re-positioned by using stage position panning because the convolution will respond by itself to that positioning as the room did. By monkeying with the stage position, you are distorting that response with a stereo source.
All this is further complicated by one other thing that bothers me - the fact that altiverb does something rather odd from the standpoint of orchestral performance. Almost all of the impulses are recorded with multiple MICROPHONE positions but only one IMPULSE position. This is exactly backwards of what is needed with convolution recordings for orchestral use. Because what you really want is one microphone position ( the listener) and multiple impulse positions (the instrument or ensemble). It was mentioned earlier in this thread that things started to sound phasey when mixing different impulses. I have not noticed that, but it makes sense that would happen, because it is as if you are hearing the same concert hall several times when you mix impulses, instead of one concert hall with different impulses in it. Or it is almost as if your head were in multiple places as you sat in the hall ! The only exception to this weirdness is the Todd AO set, which has single microphone placements with multiple impulse locations. That is perfect for orchestral use, but it is not the most beautiful reverb space unfortunately.