Resolve conflict feature is confusing/dangerous

The feature to resolve a conflict is highly confusing and dangerous. Here's why:

Let's say we have two branches, master and production. It is always safe to merge production into master but it might not be safe to merge master into production (for example, because master can contain some untranslated texts that need translation before they can be put into production). Now someone would like to merge production into master. If you do that via Git directly, a merge commit is created on the master branch which includes resolved conflicts if there are any. If that is your assumption how "conflict resolution" works and you apply that assumption to GitLab's feature of "conflict resolution", very bad things happen. Because GitLab merges master into production, then applies the conflicts that were resolved in the UI, and commits the result of this into production.

I'd argue this is highly confusing and dangerous. There is no indication in the UI that this will happen (unless you have read and remembered GitLab announcements or consulted the documentation).

At the very least, the consequences of the action should be listed on the page, but in general I would say it is a bad idea to have "resolve conflict" work completely differently to how it works via Git directly.