I did a bit of this last night, using the CR, and it worked fine using a Studio send, with its inserts. I actually did it for overdubs*, rather than as an A/B, but I think it’s the same concept. Really not too mind boggling (by definition, if I am able to do it). Your suggestion sounds like it would work too.
I’d love to hear how the professional/high-volume guys do this though ( hint/hint 
).
*Overdubs using Studio Sends (using old vocal as a guide vocal, and existing piano track to sing to):
– “Old vocal” to Studio 1, panned hard left. Volume way down.
– “Old piano” to Studio 2, panned center.
– “New vocal” to Studio 3, panned center/right. A little bit of comfort reverb as an insert there, and a compressor for a little gain only. I can probably get rid of this by turning the other two tracks down and the headphone output up, but, hey, it’s a work in progress.
– “No output” in the “real mixer” (F3) for any of the above tracks.
– Cans for tracking.
Then I record the “New vocal”, which I hear loud and clear off center to the right a bit … using the “old vocal”, panned hard left and turned way down, as a guide vocal; and the piano track center and turned down as well.
So far so good.