In my experience the one certainty of modular is that once you get your modules you will realize you need something else 
I would say start small with the minimum that you find interesting and then see where you are running into limitation and then decide if/how you need to move beyond them. Get a bigger case so you can add a few more modules than your starting set.
infovore gave some great advice. I find it’s also interesting to figure out how you can use your existing modules in novel ways to perform functions that they aren’t explicitly tailored to.
