se for WLS and FIML

[Maximilian-Stefan-Ernst]

Standard errors for WLS and FIML may not be computed correctly at the moment

[Maximilian-Stefan-Ernst]

Seems to be correct now; however, the tolerance is quite high (1-e2 for WLS). Maybe ask Yves about this.

[Maximilian-Stefan-Ernst]

I revisited the computation of GLS standard errors, and it seems to be correct. I suspect the high tolerance is because the GLS formulas are slightly differently implemented.