One case for not using Wikidata for Geneology would be that specific subtrees should be nonpublic. The present system prototype that builds on MediaWiki won't be able to answer that use-case.
Another case would be the desire to allow people to upload data that's not based on any sources. If that's the desire for the new project, then there's a completely unanswered question about quality control.
So, focus on coding something independent of Wikidata?