You're right, to render a static site you don't need a database. However, for a lot of CMS functionality, a database is required (or in our case a "Data Layer"). It's used for searching, sorting, and filtering content. In Smashing Magazine's case, they need to do these operations across 10's of thousands of Markdown files. This couldn't be done without our Data Layer and would result in rate limiting if they hit the GitHub API directly.
Also, since our Data Layer provides a GraphQL API, developers can do interesting things like pull their Markdown content into other applications and use server side rendering. We see it as the best of both worlds: you own your content in Git but you also get a headless CMS to query.
Semi-related: we're on the cusp of releasing some improvements that will make self-hosting Tina much better, so check back in a week or two if you're interested in that.