prefetch_related_objects fails to cache UUID FKs when the string representation of a UUID is used

[selcuk]

Trac ticket number

ticket-35434

Branch description

Added unit test to reproduce the issue in the title.

Checklist

• This PR targets the main branch.
• The commit message is written in past tense, mentions the ticket number, and ends with a period.
• I have checked the "Has patch" ticket flag in the Trac system.
• I have added or updated relevant tests.
• I have added or updated relevant docs, including release notes if applicable.
• I have attached screenshots in both light and dark modes for any UI changes.

[github-actions]

Hello! Thank you for your contribution 💪

As it's your first contribution be sure to check out the patch review checklist.

If you're fixing a ticket from Trac make sure to set the "Has patch" flag and include a link to this PR in the ticket!

If you have any design or process questions then you can ask in the Django forum.

Welcome aboard ⛵️!

[jacobtylerwalls]

Thanks for the patch.

Had a look at whether the model with ManyToManyField could be used to test this and didn't find a way, so I think the new test model is fine.

One suggested change to clarify that the underlying functionality still works (and not merely doesn't crash).

Also verified this fixes the issue I encountered in my project (mentioned on ticket) 🚀 .

[jacobtylerwalls]

I can mark this "ready for checkin" on Trac if you can address two things from the patch review checklist:

• commit message & pr title: use ticket number and period, like: Fixed #XXXXX -- Avoided bad thing.
• release note in docs/releases/5.0.5.txt. Although it's not a regression in a recent version (I had this problem on 4.2), ObjectDoesNotExist inappropriately escaping qualifies as a "crashing bug" under my understanding of the backport policy. Someone may correct me on this point.

[sarahboyce]

Based off the ticket discussion, I think this PR won't be accepted