NBI & NBII not indexing correctly when falling back to poisson for small sigma

[joranE]

For several of the functions in the NBI & NBII families there are lines of the form:

fy[sigma<=0.0001] <-  dPO(x, mu = mu, log = log)

that fall back to a Poisson density value for small sigma. However, the lack of any indexing on x or mu on the right hand side means that you get a vector length mismatch if this actually happens. I believe in each case the code should actually be doing something more like this:

idx <- sigma <= 0.0001
fy[idx] <- dPO(x[idx],mu = mu[idx],log = log)

A crude check of the source seems to surface this mistake in BB.R, NBI.R, NBII.R, DELAPORT.R, GeneralizedPoisson.R, ZIPIG.R and NBF.R.