GeezerVison

Basic improvements to the batch processor work now, specifically that chunking happens and so there is no real limit to the input list size.

What's next:

• Add an option which if turned on, a 429 status on one of the responses doesn't discard all successful responses so far.
• Add a scaling and hard timeout, and options to override them.

From this I realized we can create an iterator-based batch function, so filed #38. The existing non-iterators can be refactored to reuse the iterator later by consuming all of the iterator at once before returning.