I've never encountered any treatment of attenuator topology, but if someone off the street asked me to choose, I'd pick T over Pi, because series R up front does a better job of hiding non-idealities than shunt R. Think about it: a leading shunt R demands perfection in the transmission line leading up to it, while a series R devours imperfections both prior and forward.

I have seen printed horns coated with the spray-on conductive paint (e.g. MG Chemicals 843AR) whose gain is only a dB below a metal one, but it was a one-off and I never saw any follow-up. That said, it might be enough for the initial strike plate needed.

As far as smoothness and surface sealing, you should look up what people are doing to ABS with acetone vapor. I've tried it a bit and can attest that, by paying careful attention to temperature and duration, you can get quite nice finishes.

My recollection about choked flanges were that they were intended to mitigate power leakage escaping from the mated flanges. But this was specifically relevant to microwave installations involving high average transmit power, e.g. more than 5 kW, where leaked power could be troublesome.