See above. If your emotional state (and I assume the ability to distinguish a state of heightened emotion from a resting state) causes you to act in ways which do not reflect your evidence-based assessments, it is causing you to act against your rational decisions and is therefore irrational.

I would say that the ability to make this judgement belongs to the best-informed rationally-acting observer: someone who has knowledge of the your mental state in both emotional states, and from the available evidence, estimate whether or not the difference in behavior can be attributed to emotional causes. This observer may very well be you yourself, in a resting state; once you have regained your perspective, as you have a lot more information on your own mental state.

To expand on the example I gave above, someone experiencing social anxiety may suddenly focus on the various ways in which a social interaction can go horribly wrong, even if these futures are not very probable. Basically, anxiety hijacks the availability heuristic, causing an overestimation of the probability of catastrophe. Because this adjustment in probability is not based in evidence (though this point could be argued), it is irrational.

This definition of "irrational emotions" does not depend on the utility function used. If someone weights failure more heavily than success, and will go home unhappy at the end of the night if they have 9 successful conversations and 1 boring dud, they are not necessarily irrational. However, if, on previous nights with substantial frequency they have gone 10 for 10, and before entering a conversation they freeze in fear - then, their expected value has changed without sufficient reason. That is irrational emotion.