I suppose it makes sense for it to render as stereo. After all, if you’re rendering a mixdown it would include not only inserts but panning and level changes at the fader. That sort of implies that we’re looking at the signal after it’s been panned to the channel count of the destination channel/output.
I think it’d be potentially confusing if we had the option of doing this batch export pre-fader for different reasons.
So this is why I’m thinking that render-in-place seems like the logical place to do what you need to do. Not only that, but I think there are other benefits for you over batch exporting files.