APIv2: duplicate UUID check is flawed after writes were added

[legoktm]

Description

        # Process events in snowflake order.
        for event in sorted(events, key=lambda e: int(e.id)):
            result = handler.process(event)
            for uuid, source in result.sources.items():
                response.sources[uuid] = source.to_api_v2() if source is not None else None
            for uuid, item in result.items.items():
                response.items[uuid] = item.to_api_v2() if item is not None else None
            response.events[result.event_id] = result.status
...

    if requested.items:
        submission_query: EagerQuery = eager_query("Submission")
        for item in submission_query.filter(
            Submission.uuid.in_(str(uuid) for uuid in requested.items)
        ):
            response.items[item.uuid] = item.to_api_v2()

        reply_query: EagerQuery = eager_query("Reply")
        for item in reply_query.filter(Reply.uuid.in_(str(uuid) for uuid in requested.items)):
            if item.uuid in response.items:
                # Fail if we get unlucky and hit a UUID collision between the
                # `Submission` and `Reply` tables.  This is vanishingly unlikely,
                # but SQLite can't enforce uniqueness between them.
                raise MultipleResultsFound(f"found {item.uuid} in both submissions and replies")
            response.items[item.uuid] = item.to_api_v2()

if an item is modified as the result of a write event, and then it is also a requested item, and a reply, then it will already be in response.items and trigger our duplicate UUID detection.