Only unintegrate after closing unfailed pull requests

[rudymatela]

~There was an instance of a rebase being triggered by a PR closure. This should not happen:~

PRs 19 and 36 were closed in that order at: 13:25:59 and 13:26:00. Seemingly, closure of PR 36 caused it to be rebased, but actually that was caused by the closure of 19:

There is still a bug here: the closure of 19 should not have triggered a rebase of 36 as it had already failed. The handlePullRequestClosure logic has to be fixed. We should only unintegrateAfter when the PR being closed is unfailedIntegrated.

(Thanks to @Gijsvs for reporting.)

TODO

• [x] add failing regression test
• [x] (think about the fact that that 36 was already failed... its approval was outdated already, shouldn't have been integrated either way... -- update: I think with just the fix of the item below this should be prevented)
• [x] only unintegrate after unfailingIntegrated PRs

[rudymatela]

Here's how to extract the relevant logs:

journalctl -S "2022-09-02 13:23" -U "2022-09-02 13:28" -uhoff >closure-triggering-rebase.log