I think this is a pretty clear fold. If you push, you are likely to get called and overcalled by the blinds, who will then check down most of the time. That means you will have to beat both to avoid bubbling out, and A5o doesn't win that often in a 3-way pot.
If you have to quantify the size of the mistake, then you have to look ahead. I think if equal short stacks are eliminated on the bubble, then they split 3rd place and get 10% each. That doesn't happen very often, though, since a short stack could win the main pot. Anyway, folding with 30 chips left will be decent, so you will need to be ahead of the pusher's range to call next hand, but you should be willing to take some gambles. I think although the ICM does not properly evaluate the situation that you post all but 30 chips in the big blind, it will reasonably well evaluate the outcomes of that hand, so the Nash calculator's EQpost can be used to quantify the value of hitting the blinds.
The ICM should slightly overvalue doubling up, since you do not become an overwhelming favorite (50:1) to make the money if you double up, just a big favorite--the short stack could survive going through the blinds all-in, and you would not have enough to post two orbits of blinds. Since doubling up does not gain that much, you need to be more of a favorite over a random hand to push.
It still looks like it would take a lot more work to quantify how right it is to fold, but consider this: Are you a favorite to make the money if you fold? Clearly yes. Are you a favorite to make the money if you push? I don't think so. Since so much of your equity comes from making the money, this is the dominant factor.