That tinkertoy approach actually costs you more in the long run. Most missions we have now don't need assembly. It is just better to have all liquid handling done here on the ground. ISS modules are rather cramped--especially the Soviet versions. Remember, the American ISS parts werelaunched by shuttle so they are just cans--all propulsion was handled by orbiters which no longer fly.
A couple other folks on SLS--Carolyn Porco of the Cassini mission
So, yes, the capability of the rocket DOES come before the mission design and the payload determination.
The other is the size of the SLS. If down the road we wanted to launch a telescope that could, for instance, study the entire energy balance of Earth with pixel sizes smaller than clouds, it would take a big telescope. With a big rocket, you can think start to think about launching big optical systems. We think of the SLS as the human spaceflight programme, but it could be hugely enabling for science.