Dicke.liouvillian() not to be used for the spectral properties or nonlinear functions on steadystate

[nathanshammah]

The spectrum of the Liouvillian obtained with in the permutational symmetric basis from Dicke.liouvillian() or Dicke.lindbladian() is not to be used to look at spectral properties, such as the Liouvillian gap and overall spectrum.

Also, nonlinear functions such as the entropy of the steadystate can be wrong. Any test on expected values of collective spin operators the steadystate, compared with the steadystate calculated in the full basis, give correct results.

Working on a fix and clear disclaimer meanwhile.