When you say "effectively uncertain outcome", do you include situations where the events are random but the odds are predictable?
Let me give you an example. On average, there are 14 storms big enough to be named in the 6 month long Atlantic hurricane season. If a prediction market was saying 30% odds of a storm big enough to be named every day for all 183 days, betting against that would be free money. Would you call it gambling to make the same bet on all 183 days? The day-by-day outcome is uncertain, but the overall outcome is extremely certain.
Yes, I would call it gambling simply because someone has to take the other side of the bet and lose.
The entire point of there being a gambling "line" is because two parties have to agree on a wager that they both think has positive EV. That's effectively gambling. Somebody has to lose for the other party to win.
Obviously if the counter-party is an institution with a legitimate need to hedge, it becomes an insurance policy, but that is a world of difference than just two counterparties wanting to make bets for fun.
I think you're close to a good metric, but you need to consider the situations where one person doesn't have a positive EV expectation, or where that expectation is provably wrong. I think those situations can empower a winning non-gambling actor.
One participant in a market can be gambling while another participant isn't gambling. In particular, casinos don't gamble.
Also for many things there exists a scale from fully random to fully skill-based. So in my opinion things can be semi-gambling with a lot of gray area.
Just calling it gambling emphasizes the former problem while dismissing the latter problem.