- estfun, bread, vcovHC
- fix issue in profile.gnm (Matthias Pohlig)
- adapt DrefWeights and getContrasts to accept alternative vcov

BUG in anova.gnm

e.g.

rc1 <- gnm(Freq ~ r + c + Diag(r,c) + MultHomog(r, c), 
           constrain = "MultHomog(r, c)SM",
           family = poisson, data = friend, verbose = FALSE)

anova(rc1)

because constraining one element of 4th term, treats as if all constrained!! (Although shouldn't have to refit full model anyway!)
