Semantically Consistent Video Inpainting with Conditional Diffusion Models