That's a cool website. I would say my usecase is really focused on picking existing n tables in a db for like a design or presentation. I would not want to handcraft tables as I don't care what the diagram looks like; all I care is that the diagram accurately reflects the schema.

I intend to stick true to my inspiration and keep dbdot focused for that usecase. Hence, I expect this tool to actually reach a completion after which I don't think I will want to add any more features.

That's not high level design though. High level would be (for example) in Barker-notation were you have subclasses and n:m relationships are not represented as their own entities.

Only as a next step you translate that to the actual relations.