Move back to using DEfaultThreadPools for reconnect tasks - as for very bad connections, DedicatedTaskRunner could get swamped with reconnect tasks