I agree that this would be ideal, but for practical reasons iNaturalist changed this about 4 years ago so that it was an option (not required) to retain the input taxon of a split as one of the output taxa. See this blog post for the full explanation (you have to scroll most of the way to the bottom to find it).