hi Jakob,
I've added an answer to the so post, and I also noted this kind of interaction isn't fully declarative yet.
I have had a similar case with open xml xlsx documents where I wanted the rows lazily processed but everything else copied over (strip extra blank columns and rows before loading in poi). That example can work with normal filter ops, but its not as obvious as either the example on SO I posted or the others on the iteratee examples.
Re more examples, I'm happy to link to the SO post in the docs for 0.5.0. And when I've figured out a nice way to handle lazy loading of "some bits" then I'll add docs for those too :)
Cheers,
Chris