The previous commit landed the doc and the probes but its BACKLOG edit asserted
on text that had since moved, so the index did not mention the hunt at all.
That is the third time an assert-then-replace patch has silently failed this
way. The pattern is the problem: the assertion aborts the edit but the commit
that follows still succeeds, so a missing index entry looks like a clean run.
Locating the anchor first and verifying the insert afterwards, as done here, is
the habit to keep.
462c3dc019