Accept the usual user settings related to file formats.
Most of the diff are the mechanistic code changes required to allow
providing the required FormatSettings to the format factory. The File
engine then extracts these settings from the `CREATE` query, and specifies
them when creating the format parser.
This will just pollute the output log.
And this is significant for kafka streaming, since it will print this
message for each message (due to how
ReadBufferFromKafkaConsumer/KafkaBlockInputStream works).
You can see this in the server log of the test_storage_kafka
(integration test), for the table with kafka_skip_broken_messages > 0,
it will have 52 lines of such message, 2 with "Skipped 1 rows" and 50
with "Skipped 0 rows" (since there are 52 messages in summary, 2
malformed and 50 correct).
