CDC-to-Kafka 4.3.0

This release speeds up Avro serialization significantly, leading to a 2-3x increase in overall throughput. It also includes a few minor bugfixes, namely:

1. Field truncation (when configured) now properly operates against encoded bytelength rather than string character length
2. A fix was made for failures to skip unnecessary new snapshots whenever a DDL statement which added a new nullable column to a tracked table used bracket-quoting for the new column's name
3. The show_snapshot_history tool no longer raises an exception when queried for a topic that does not yet exist