This seems so obviously useful and surprising it's taken industry so long to get here. But honestly I'm hesitant to go for something that isn't like a wrapper around the underlying database, just because those are notoriously difficult to implement correctly. Postgres has been around a long time, and has been battle tested, etc.
Yes, it's definitely an uphill battle to launch a new DB for this reason. Our current customers are people who need version control bad enough they're willing to take a chance on a younger implementation.
That said, we have 99.99% correctness according to sqllogictest: