>To start my work, i decided consider only HOMO-1, HOMO, 2xLUMO orbitals in my active space. So, i had done SA-CASSCF(4,4)/6-311G(d,p) with nstate=6
In CAS(4,4), the ground-state configuration is 2200, and singly excited configurations are 2110, 2101, 1210, 1201. Others are higher excitations. You ordered nstate=6 and expected that all 4 single excitations are among the ordered states. Obviously, it was not the case (examine your CAS-CI results). Probably, some double excitations intruded instead of desired single excitation, and state-tracking option did not allow single excitation to get back. So, order more states (say, nstate=9), find all single excitations and include them in wstate (something like wstate(1)=1,1,1,0,0,1,1,0,0,0) with state tracking. Do 5-7 MCSCF iterations and check the order of states. I am sure there will be no intruders any more (all single excitations will be lower than doubles), and wstate becomes 1,1,1,1,1,0 as expected. With this, you may finish the calculation. The energies of all states of interest will be better, too.
After that, do XMCQDPT.