Some remarks about the two-matrix Penner model and the Kazakov-Migdal model