Heart rate, respiration, O2 blood saturation, skin conduction (okay, I'm not saying I know what I use it for, but it would be fun to play with - back in my teens this is one of the things we used for a biofeedback mouse...) I wonder if you could pull out a good proxy for hydration? You'd probably have to tune it to the individual... Blood pressure would be neat, but difficult, I think. (At least compared to how it is conventionally taken.)
I'd be tempted to build an MP3 player into while I was at it. Always a bit of a pain to haul tunes around while running. (Of course, really, I want data connectivity and maybe a little more brains, but that might be superfluous. Some of my sports bras already have a pocket for my phone.)
On a purely personal level, I have no lack of cup space - there were some weird hormone fluctuations during the whole spine injury thing, and at the end of things I lost the weight I'd gained, but not the extra cup sizes, and while they're kind of annoying not so much so that I've had them reduced, yet. (There are a lot of active women with larger breasts. And we spend a fortune on bras. Gods, you do not want to do spinning flying jump kicks without having your breasts constrained. Running is also especially fun, and it's pretty common to wear a compression style sports bra over an encapsulation style one.) But I don't think we're talking particularly high power consumption, and I don't think putting a battery in the cups would be my first choice. Give the bra a nice wide strap in the back - which will make it more comfortable anyway, and put a pocket for the battery there. Something along the lines of a cell phone battery should to it, and you have fewer nerve endings on your back and spines curve less backwards than forwards, so you're less likely to end up with it pinching. And you could easily fit two - and smaller separate batteries will be more comfortable than larger single ones.
Of course, there still is the question of whether we've talking about multiple bras that can take the sensors, or sensors in some kind of framework that easily and comfortably inserts or attaches to bras. The latter might be more difficult to do well, but I'd tend to prefer it.
There's a lot of variability in terms of what is expanding during breathing - I think one strain gauge, and figure you're not going to get a lot of data, or go to the other extreme and outfit a close fitting tank. I find myself with mixed feelings about an accelerometer - I do soft body biomechanics as part of my academic research, and while I'm not doing motion capture as such, a lot is done in my field, so I have an immediate response of - oh yeah, we'll find something to do with it! But then, I also do soft body modelling, and use community tools when I can (OMG, trying to model slug mouths in bullet blows up so very badly...) which means, of course, that I'm well aware of the vast amount of time and effort that has gone into beter modelling of breasts for the gaming market and... yeah, no. (I think, on the whole, the opportunity for movement analysis is worthwhile, and you'd have to include quite a few for breast modelling. But then, these days accelerometers are cheap.)