Thanks for your attention on this.
Based on the source I think it's likely that 2.13 behaves the same as 2.7.1. According to the PR linked from this report, the fix was merged to the jenkins-2.0 branch but there's a comment that might indicate that was a mistake: https://github.com/jenkinsci/packaging/pull/51#issue-144292301 says "A separate jenkins-2.0 packaging branch has been created, because pre-Jenkins 2 users may rely on AJP." which to me indicates that the fix should have been applied to master but not jenkins-2.0 (since the fix disables AJP). The fix was applied to jenkins-2.0, though.
It also looks as if the 2.7.1 RPM's are built with the master branch of jenkinsci/packaging instead of the jenkins-2.0 branch, as Daniel indicates. I supposed if both of these things were fixed then the result would be no change to the current behavior. Unfortunately, seomone's going to feel some pain, I guess, either the set of those using AJP, or the set of all new and upgrading users.
Would it make sense to re-open this report to track the effort to get the fix to end-users, or is there a different report to track that?