Some remarks on game theoretical approach to prediction of formation of parliamentary coalitions. The case of Polish elections of 1997