That definitely sounds like a bug in the product that is difficult for you to address with less than changing the server and database collation.
It seems that DevOps in some situation for some reason wants to enforce a collation that is accent- and kana-insensitive collation, so it overrides your server collation, but in the wrong way. There are no collations with KI in the name, but kana-insensitive collations simply recognised by the absence of KS in the name. That is, the correct collation name is Latin1_General_CI_AI_WS.
My expertise is with SQL Server itself, and I am out of touch with Azure DevOps these days. I'm a little surprised that a bug like this would go unnoticed for four years, assuming the DevOps server 2022.2 was actually released in 2022. While kana-sensitive collations are not that common, they are not so rare that you would be the first one to run into this issue. So I would check if there are later versions/patches that addresses this problem. If you can't find any patch, open a support case with Microsoft.
I would not recommend that you change the server and database collation, because this is a very complex and time-consuming operation. But as it happens I have an article that describes how to do this: https://www.sommarskog.se/collation-change.html And, of course you may have good reason to use a kana- and witdh-sensitive collation. Note however, that these settings only applies to data in Japanese.