APIPaginator is used by hosting services to more easily work with

paginated data from a hosting service, taking care of managing page

data, previous/next URLs, and HTTP requests.
This change enhances the support by taking advantage of the new query=

support for hosting service HTTP requests, allowing callers to more

easily customize their requests in a way that applies to all page

requests, not just the first. This also means the paginator gets to

leverage the cross-Python compatibility already present in the hosting

service's URL building.
New iteration helpers have been added that make it trivial to iterate

through pages of information, or all items across pages. This can be

capped to a maximum number of pages, helping ensure that callers don't

iterate forever (for instance, when iterating through the commit tree in

a repository).
This does change the result of a single GitHub URL, but it's harmless

and only impacts tests. This is due to the ordering code not kicking in,

since it's a fully manually-built URL. A separate change will

standardize URL building for GitHub.

Unit tests pass.
Tested this with some upcoming code that's taking advantage of the new

URL building and iteration helpers.