Bug: if there was a timeout when calculating the page count, the timeout
wouldn't be cached, so we would hit multiple timeouts trying to
calculate the page count multiple times.
Fix an invalid SQL exception that occurs when the paginator tries to
do a COUNT(*) to calculate the page count of a relation that already
includes a GROUP BY + COUNT(*) clause. We need to nest the whole query
inside a `SELECT COUNT(*) FROM (...)` subquery so the inner COUNT(*)
doesn't mess up the outer COUNT(*).
Fixes#4285.
Add <link rel="prev"> and <link rel="next"> elements to most pages with
pagination. This should work on all index pages, but it won't work for
things like pool or forum topic show pages.
Also remove the <link rel="top"> element (wasn't useful, was just a link
back to the root url).
Previously the page-based (numbered) paginator would always count the
total_pages, even in API calls when it wasn't needed. This could be very
slow in some cases. Refactor so that total_pages isn't calculated unless
it's called.
While we're at it, refactor to condense all the sequential vs. numbered
pagination logic into one module. This incidentally fixes a couple more
bugs:
* "page=b0" returned all pages rather than nothing.
* Bad parameters like "page=blaha123" and "page=a123blah" were accepted.
