Is there any improving the test suite to catch this kind of thing with the refleak bots? A quick glance suggests probably not, but it's worth at least getting a proper answer to the question written down :)

I think a few of these could be caught with a test, yes. It doesn't feel extremely high value (these particular paths aren't likely to regress), but I can add some if you like.

I'm in no position to make that value judgement :). More test coverage is generally better, but if they're not worth it they're not worth it.

Thanks @JelleZijlstra for the PR, and @encukou for merging it 🌮🎉.. I'm working now to backport this PR to: 3.13, 3.14.
🐍🍒⛏🤖

Sorry, @JelleZijlstra and @encukou, I could not cleanly backport this to 3.14 due to a conflict.
Please backport using cherry_picker on command line.

cherry_picker bd13cc09faaef01635aea85130f33aa8cbb8b177 3.14

Sorry, @JelleZijlstra and @encukou, I could not cleanly backport this to 3.13 due to a conflict.
Please backport using cherry_picker on command line.

cherry_picker bd13cc09faaef01635aea85130f33aa8cbb8b177 3.13

GH-145712 is a backport of this pull request to the 3.14 branch.

GH-148661 is a backport of this pull request to the 3.13 branch.