On Sun, 12 Apr 2020 17:35:31 +0300
Eli Zaretskii wrote:

> Let's just come up with a local change in gnus-sum, for the emacs-27
> branch, okay?

OK. I think the pat(c)h of least resistence would be as follows
(everything just the same as for master, but with the helper function
localized to gnus-sum and no changes to other files):