I think a lot of this is simply inaccurate. E.g., type() can be used in the select-clause, we have tests doing that. As for how type() is ultimately rendered in SQL.. that actually depends on how you have defined inheritance; for joined-inheritance, yes, it will use a CASE statement. One improvement there would be to leverage the added ability to combine joined-inheritance with an explicit discriminator column.

Either way, 4.x is no longer maintained. If you are willing to try these assertions against 5.0 then I can look at some solutions.