I wonder if what we're seeing is a tube that was not fully inserted into the joint prior to brazing, leaving the lug to support the stress at the joint, with a crack appearing at the point where the end of the tube lies under the lug?