I've looked into Dolt perviously for powering our product that is Git for financials.
The trouble is that most of our complexity lies in our data's relationships. Merging is very tricky when some data that has been added in one branch has not had a specific change in properties when other data in a master branch has been modified.
Good point - and why a graph based approach has advantages in some use cases (disclaimer TerminusDB co-founder here). Allows more flexibility in versioning relationships and more scope to merge in the scenario you outline.
The trouble is that most of our complexity lies in our data's relationships. Merging is very tricky when some data that has been added in one branch has not had a specific change in properties when other data in a master branch has been modified.