The problem of boundary-layer flow encountering a three-dimensional hump revisited