While brainless did commit the survivorship bias, I would consider you responding to what he said rather than what he meant to be attacking a strawman.

In this case, lets consider all humans who ever considered suicide. Now, while it's true that we can only observe those who did not commit suicide, we do have a substantial amount of data on humans in general that we can work from. If this claim has you skeptical (ie "successful suicides are a separate reference class"), I will remind you we have data on suicidal people who were forcibly restrained, which you are encouraged to look up.

So now let's look at the core of brainless' argument: people change with time. If you look at happiness rates for quadriplegic patients (or any of the hedonic treadmill research), you'll see that your happiness levels aren't so impacted by life events. This effect is present in both the population of completed suicides, and aborted suicides, and is likely a lot of the causal force brainless is trying to reference.

Which is to say that it's highly likely that people who kill themselves counterfactually wish they didn't.