We have functions in HostingService subclasses that need to deal with

paginated data. Right now, they're all handling it in their own ways,

but as we move to wrapping more of these in the web API, it's helpful to

standardize them.
This introduces a set of paginators that can be used and subclassed by

the more specific HostingServices to provide a standardized API for

pagination of data.
APIPaginator is designed to perform API requests and handle pagination

through links returned through the payload or headers in some form.

Subclasses are expected to define how API requests are made and parsed.
ProxyPaginator attaches to another paginator (typically an APIPaginator)

and transforms the results. This allows, for instance, a

HostingServiceClient to return an APIPaginator with raw results, and a

HostingService to wrap it and return structured data derived from those

raw results.

Unit tests passed.
Tested this manually with an accompanying change to GitHub's

get_remote_repositories. I was able to page back and forth and get the

results. I also tested start= and per_page= with that.