The generation process is the one you quote. On the other hand, my original question was “what is the approximate exponent of the degree distribution of the sampled networks”. I think I was clear about the detail I need. For this question, you may contrast the generative process described in the documentation with existing literature, or make aggregate statistics, etc but the documentation has no direct answer. If the answer comes from the only model cited in the documentation, which is only one of the infinitely diverse models you may produce with this function, the exponent should be 3. But one knows from the literature that the distinct assumptions of the generative process that function sample_pa() implements, generate asymptotic distributions with different characteristics, in particular, with different resulting exponents. That is the motivation of my questions, regarding which estimates from the literature are valid for the processes implemented by this function.
Since the documentation has no direct answer, I put the question on this forum. And I do think that putting only 1 reference in the documentation about only 1 of the possible models you may get, may be misleading about the networks you are working with.