Today, in our current setup, it happened that an accepted diff was closed by a commit. However, at the same time, another developer accepted the diff.
Presumable because of a race condition in the flow of actions, this caused the diff to be re-opened in an accepted state.
Attached a screenshot of the header and timeline of the diff in the final state.
We didn't manage to reproduce, only to produce, see the attached screenshot. This on its own is enough to identify that the bug exists, but since this is probably the cause of a race condition, reproducing is not very deterministic.