On the possibility of two-dimensional growth of GaAs on atomically flat Si(100) surfaces