I tried to track the label problem down yesterday, I might give it another try in case it would only be a UI change. I suspect that it could be, because find-usages does seem to do a better job to find the name of the enum constants than call-hierarchy - while both use the same code path to compute the usages.

i also forgot to mention that i can't access the old bugzilla atm. I asked on slack and it seems like more can't log in. I would also be good to read through the bug IDs which were mentioned in the comments marked as hotfix. I might open a infra issue (edit: INFRA-27390).

i can now view the bugzilla bugs again since my account got repaired, the comment mentions the hotfix for https://bz.apache.org/netbeans/show_bug.cgi?id=213723 which is the exact same bug.
reproducer was:

public enum Tester {
    ENUM1 {
        {
            System.out.println("");
        }
    },
    ENUM2 {
        @Override
        public String toString() {
            return super.toString();
        }
    };
    private Tester() {
    }
}

what I believe happened is that getStartPosition() used to return -1 for enum fields with "class bodies", but at some point it began returning positive start positions, while getStartPosition() still returns -1 to this day. This made the same bug resurface again.

So this PR essentially activates the old hotfix again. cc @jlahoda