Can't help you with the setup, but here is my understanding of each component. RVV = Rail Vehicle View. It's the visual aspect behind a rail vehicle. This includes but is not limited to: static meshes, skeletal meshes, and all the interactible knobs/switches. RVM = Rail Vehicle Model. This is the physics/simulation side of things. The RVM implements the Simugraph and has programming for all the interactible knobs/switches.
The RVD links together the View and model actor and some other stuff. When your make a formation node it's the RVD which is put in for each vehicle entry.
I wrote a short and simplified guide on that, but maybe it can help you nonetheless: It doesn't cover multiple units yet, as their setup is not quite as simple, but it tells you about the basics
If you want to understand the purpose of the separation between a "View" component and a "Model" component go read up on the standard software development technique of "Model View Controller".
I just noticed this post and wanted to say it looks quite helpful! I'll be looking it over this weekend for sure! I look forward to eventually learning about creating multiple units, too!