Hmm, this doesn't appear to be a regression after a bit of digging, the associated code (includes/Hooks.php:227-231) seems to have been last changed 4 years ago.

Change #1059928 had a related patch set uploaded (by Sohom Datta; author: Sohom Datta):

[mediawiki/extensions/PageTriage@master] Reset reviewed flag when article is redirected

https://gerrit.wikimedia.org/r/1059928

Auto unpatrolling redirects to articles makes a ton of sense, since this closes a UPE loophole and a notability check loophole.

Does auto unpatrolling articles to redirects makes as much sense? There is no UPE component. The most likely use cases that come to mind are vandalism, and BLARs (including BLAR disagreements between two editors / BLAR edit wars).

This isn't a firm "no" from me, but I do think we should talk about this a little bit.

Cons: Can our normal anti-vandalism processes handle this without needing to involve NPPs? Do we want to be inserting NPPs into BLAR edit wars?

Pros: I suppose we should be checking "every redirect" from an RFD perspective, and this appears to be a loophole of redirects that are not currently getting checked.

@Hey_man_im_josh on Discord says this is a Regression, so let's go ahead and fix.

Sounds like we don't really understand the root cause though?

As long as we confirm it doesn't work in localhost pre-patch, and does work in localhost post-patch, should be OK to skip figuring out the root cause.

We will take a look at it as soon as tests are added to the patch!

Change #1059928 merged by jenkins-bot:

[mediawiki/extensions/PageTriage@master] Reset reviewed flag when article is redirected

https://gerrit.wikimedia.org/r/1059928

This is not a regression, as there was never a time when articles converted into redirects were added to the PageTriage queue, as far as I could tell.