Ah I see, thanks for the explanation. In our case, the nested schema is neither fully known nor fully consistent (I'd say mostly consistent, but there are edge-cases). I'll need to tinker with this a bit, but I think I have a workaround in mind for this case.

This is a bit of an unfortunate limitation of schemaHints at the moment, but I agree it would be great if they enhance schemaHints to support this in the future!