People who initially win in gambling will kick around money for longer than those who lose on their first play.. So excluding immediate dropouts can make a gamble look like an investment.
I think survivorship bias indicates a segment that survives at a higher rate due to an actual trait and then its trait is presumed to have been a norm at the start. Survivorship Bias would imply there is an actual factor beyond gambling and we are just too optimistic about its frequency.
More simply, once you exclude people who only bet once you are excluding a disproportionate number of negative bets simply because betting a second time is based on the first outcome. If you want to do exclusions you have to exclude the outcome of the 1st n bets even from players who keep playing.
But then you get something that is also misleading as players can get more and more conservative on the way out the door. I.e. taking guaranteed pennies in splits and buyouts as they realize their big payouts from roller-coasters were not as certain as they thought. There should be a strategy that does similar to the return from a safe investment, so that also isn't interesting.