Survivorship bias as an argument doesn't really work because you are already presupposing you are the one who survived. Of course if you assume that there is a multiverse of infinite copies of yourself and at least one of them survived an incredibly incredibly unlikely event, then by definition you would not die and would be the person who survives the event.

But it's kind of circular. You cannot apply surviroship bias prior to conducting the experiment because you have no reason to believe that what you call "you" would be one of the survivors. It is much more likely, even if we assume the multiverse theory is true (see my criticism of it here) that what you would call "you" after the splitting of worlds would not be one of the survivors.

Let me give an analogy. Replace the very likely event of dying with something else, like losing the lottery. At least one branch of the multiverse you would win the lottery. Yes, if we bias it so we only consider the branch where you win the lottery, then by definition you are guaranteed to win the lottery if you play it. But that biasing makes no sense prior to actually playing the lottery. It is much more likely what you call "you" after you play the lottery would be one that sees themselves as having lost the lottery.