This is the distinction that keeps the audit trail honest. A question is part of the template — the immutable paper
the matter opened on, pinned by version at the moment you opened it. An answer is the client's response, which you
can correct through the matter's life without disturbing the paper. So when you send an answer back and it is
re-collected, three things hold at once: only the flagged answers change, the template version stays exactly what it
was, and the journal records who corrected what, when, against which version. From the command line the same loop is:
navigator notation request-changes <id> --question person__client --note "confirm the spelling"
navigator notation update <id> --answer person__client="Libra Jones"
navigator notation approve <id>
The matter you approve is the matter you reviewed — nothing re-rendered behind your back.