Better bounds for the worst case of the stretch-length tradeoff in geometric networks