I am on the right track in thinking that using the discriminator approach could be adapted to the single-db, single-schema multi-tenant use case?
(Alternatively, wondering whether @Filter might be usable for it.)