Is the relativistic contraction of bond lengths an orbital-contraction effect?