After efords comment JuliaApproxInference/LikelihoodfreeInference.jl#5 I thought it may be a good idea to move the priors to Distributions.jl. There is also this discussion in discourse. What do you think?

It is a very good idea, but i think that it will seriously hinder the speed at which we can create a workable solution.
For example Distributions.jl already has a factored distribution (it is called Product), but it doesnt support mixing continuous and discrete distributions, and ofcourse since Product is based on arrays, it is not possible to extend it and have it type stable at the same time. Thus i know for sure that Distributions is not going to accept a PR with Factored inside

i will merge this, just to start working on it, we can excise this code anytime ;)

Yes. This makes total sense. We can put whatever we want here for now and move it to other packages anytime later.