Add MaxScale Configuration Support for slave_connections_needed_for_purge Similar to enforce_read_only_slaves

Currently, MaxScale provides the enforce_read_only_slaves configuration option to automatically enforce the correct read_only state on backend servers based on their role (primary vs replicas). However, there is no similar mechanism to manage the slave_connections_needed_for_purge variable consistently across the topology.

In environments where binary log purging behavior must be controlled, it is desirable to ensure that only the primary node has slave_connections_needed_for_purge=1, while all replica nodes have slave_connections_needed_for_purge=0