??so does diaconescu's theorem (or something...) imply (or something...) that the classical model category of a presheaf topos is the _free_ filteredly cocomplete category on the opposite of the site?? ... ??something about this as maybe giving clear examples of free filteredly cocomplete category on a small category being non-small??? ... ??and also maybe supporting idea about ... ??finite colimits beck-distributing over filtered ones ?? ...???or something ?? ...