It seems that we should revisit and check how signals (and incremental snapshotting) works for multitask connectors as they tend to exhibit certain issues

• JMX signal channel should be fixed by DBZ-8137

• Kafka signal channel – since all task share the same topic and the same consumer group id, this esentially creates a race condition – the first task to read the signal wins.

• Source signal channel and incremental snapshotting). Signal.data.collection is configured per connector and has the format database.schema.table however this means that incremental snapshots would attempt to write into the same table from multiple tasks. Same with signals.... Should this be a signal table per database? It probably depends on how multitasking is done