Tell me if I’m misunderstanding, but I’m thinking that (1.) is the same as Owen’s OP request, (2.) is the same as his other recent request, (3.) is already (partially*) possible by slicing the SearchIterator, and (4.) is premised on something we’re not sure about yet.
If that’s right, I guess it might make sense to post (4.) as a separate FR? (As an alternative to combining these FRs, we can at least link between them?)
* I guess slicing would return the maximum number of rows all at once rather than batching them lazily until the maximum is reached. Is that the distinction you’re making? I’m having trouble imagining a use case for that, but then I’m a database noob.