I don't have a definitive answer, but what happens if the bot is programmed to stop at a noindex rule and not parse the rest of the document? Technically-speaking would this be wrong? I don't think so. So with your example, only testing will tell. I don't believe that either behavior would be "wrong".
I was wondering if one takes precedence over the other and if the cascade does not apply in this instance? Test? Arrrggghhh! Let me add to the other 100+ tests that may be going. Why can't this be in writing somewhere already? ;)
Or rather, in practice Google don't. But what about the others? Live, Yahoo, etc.? As there is no real, formal "standard" either for robots.txt or for meta robots elements, the above example would be open to interpretation which could differ between bots.