Profiled this. Using first_value instead of the additional sub-select made this perform slower at a factor of about 2x for arrays of 1M events.

Interesting, and very surprising! I'm going to dig around and see if I can't figure out why this is the case.

I would have guessed there would be a minor performance improvement, because the first_value approach effectively allows PostgreSQL to discard the data sooner, instead of materializing it in an additional subquery.