While you can use modeling tools in the assembly room, conforming clothes models will still "snap" back to their default OBJ position. (No morphs, no bends)
The trick is to visualize where poke-through is occurring in the posed position, and find those areas in the default position.
I agree that being able to do vertex-level modeling on clothing after it's conformed would be an invaluable feature.
I'm pretty sure the limitations have something to do with how Carrara reads Poser content?