Thanks for the detailed answer! It makes sense.
I’m pretty new to the Bayes Net side too, but it’s nice to see that the difference btw the Bayes Nets packages and PyMC/Stan is not one of concepts but one of implementation and focus, as you said. Should make going back btw the two quite easy 
BTW, I think your work and all these topics are super interesting and would make for a very good podcast episode – if you feel like it, I’d be happy to host you!