Sorry it's taken me so long to post my own thoughts here.
I use M-S a lot. I'll almost always use it as the main stereo pick-up on an acoustic grand, although I'll use one or two more mics to broaden/widen the sound. And I'll use it on acoustic guitar.
One such use, Lyle Lovett's "Joshua Judges Ruth", got pretty good response, and a smattering of criticism. I used an AKG C-24 with Stephen Paul 0.7u diaphrams oriented vertically. I always print L-R (which makes it easier to deal with through a project) but almost always accentuate the "S" by 6dB or more, which gives me a really wide stereo space - if I want less "S", I pan the stereo in, same thing. The criticism comes from the fact that Lyle sang everything live, and his voice is (depending on the tune and the intensity of the guitar) more or less always there in the guitar mic. And (if one cares) the voice is much louder in one side than the other, and (again, if you care) since I'm hyping the difference between the channels, the "other" side is out-of-phase.
Yes, the sound stage is shifted by the effect. Yes, the timbre of the voice is different left to right. No, I don't think the problems outweigh the advantages.
George