When there are many wildcard (MQTT) subscriptions, searching for matching addresses in org.apache.activemq.artemis.core.postoffice.impl.WildcardAddressManager::.addAndUpdateAddressMap can take too long, causing the associated transport thread to miss heartbeats resulting in a closed connection. The method is synchronized, so having several threads queued on this method can easily result in timeouts. Is there a possibility of off-loading this processing to avoid blocking the connection thread and / or adding thread-safe concurrent access to the underlying map / data structures?