Exclusion of self evaluations in peer ratings: monotonicity versus unanimity on finitely restricted domains