Absence of sign problem in the (saddle point approximation of the) nilpotency expansion of QCD at finite chemical potential