feat(alerts): Set up prune_old_open_period_activity

[kcons]

Our backlog is only 17m or so rows, but by reusing a task pattern we've already used before we should be able to safely get that quickly cleaned up in all regions.

A job or post-deploy migration would also be an option, but this is no harder and more or less guarantees that next time we look it'll be dealt with.

[wedamija]

Does this cleanup need to happen in self hosted?

[kcons]

Does this cleanup need to happen in self hosted?

Depends on what we mean by "need". Self hosted would also have a backlog that needs to be cleared, so this will do useful work, but if self-hosted also runs incinerator stuff (I honestly don't know) chances are the backlog is small enough that incinerator could handle it.

[wedamija]

Does this cleanup need to happen in self hosted?

Depends on what we mean by "need". Self hosted would also have a backlog that needs to be cleared, so this will do useful work, but if self-hosted also runs incinerator stuff (I honestly don't know) chances are the backlog is small enough that incinerator could handle it.

Yeah, if incinerator will handle it then we don't need to worry about it. Mostly I was wondering if we should use a migration so that it runs explicitly, because I assume we won't keep this task around forever, and so not every self-hosted install will end up seeing it.