ksql.
In both scenarios, they will represent the message key which kinda makes sense when you think about how Kafka handles "tables".
But there is also a difference between those 2: in the second scenario, we see the KEY field more as an optimization technique. As you can notice, the same field can be found in the message value, so it will be extracted from there.
Quoting from the official documentation page:
If the Kafka message key is also present as a field/column in the Kafka message value, you may set this property to associate the corresponding field/column with the implicit ROWKEY column (message key)
Regarding your second question, if you check the latest version of the documentation you'll notice that the second query is no longer valid.