Comment on "Self-diffusion in silicon: similarity between the properties of native point defects"