I've been documenting a lot with Jupyter notebook because it uses simple markdown (so it does simple formatting and images). It also executes code -- for example, I find myself plotting s-parameters using scikit-rf a lot. It can be exported to a single HTML document for sharing with others.

Seems like a lumped element solution might be possible at 1.5 GHz. Maybe "<5 dB insertion loss" means that the 5 dB in addition to 6 dB loss from the 4-way split.

Yes, I think that is correct. Your transformer will be made up of transmission lines with impedances of 50.459, ..., 297.217 ohms. I believe an impedance of 297 ohms might be hard to realize in microstrip or stripline. Maybe it has to be twin line?