Fix the Bug: CDC Outbox Event Router

A NULL payload column on an outbox row means "this aggregate was deleted" -- Debezium's outbox convention routes that as a TOMBSTONE (a record with the aggregate's key but value=None), which tells Kafka's log compaction it can eventually drop every earlier record under that key. Right now, deleting an aggregate silently produces a NORMAL event with a None value instead of a properly flagged tombstone -- downstream consumers that specifically check for tombstones (to react to a deletion, say) never see it as one.

Your task: find the bug in route()'s tombstone check and fix it. The topic derivation and the key (aggregateid, not the outbox row's own id -- getting that part right is what keeps every event for one aggregate on the same partition, in order) are already correct -- don't change them.

Example that currently fails

router = OutboxEventRouter()
row = {"id": "evt-3", "aggregatetype": "Order", "aggregateid": "order-42",
       "type": "OrderDeleted", "payload": None}
_, _, value, is_tombstone = router.route(row)
is_tombstone   # should be True
               # currently: False (a None payload isn't caught at all --
               # it slips through the same path as a normal event)

Real-world grounding

Debezium's outbox event router SMT: debezium-connect-plugins/src/main/java/io/debezium/transforms/outbox/EventRouter.java

Real Debezium's outbox router additionally supports promoting extra outbox columns into Kafka message headers and configuring the topic naming pattern -- but "NULL payload becomes a same-key tombstone" is the specific mechanism this bug breaks.