Galera using parallel replication does not retry failed event

Slave could encounter errors during parallel replication for example when an archiving job runs TOI DDL locally on the cluster while the replica applies INSERTs into the same table, the replica transactions fail certification (wsrep_local_cert_failures goes up, bf_aborts stays 0) and the SQL thread stops instead of retrying. With slave_parallel_threads=4 we get "Parallel slave worker failed at wsrep_after_statement() hook"

This then leads SQL-thread to stop

[ERROR] Slave SQL: Could not execute Write_rows_v1 event on xxx; Duplicate entry 'xxx' for key 'PRIMARY', Error_code: 1062; handler error HA_ERR_FOUND_DUPP_KEY; the event's master log mariadb-bin.123456, end_log_pos 1234, Gtid 1-1-1-1, Internal MariaDB error code: 1062

[Warning] Slave: Duplicate entry 'xxx' for key 'PRIMARY' Error_code: 1062