Since original GTID will be forwarded to all nodes (MDEV-20720),
such feature should be possible.

Given a setup like
https://mariadb.com/kb/en/configuring-mariadb-replication-between-two-mariadb-galera-clusters/

Also this setup with additional async replciation between the nodes.

A workaround is to establish a maxscale readconnrouter between both galera cluster,
with router_options=master, but this will only affect an outage of the master node, not slave node.

So feature request here is a mechanism, that Maxscale check async replication between both galera cluster and switch slave and master node in case of a failure
and take into account, that binlog position can be different and can determine the correct GTID.