fix: Keep pagination iterators advancing past fully-filtered pages (#964)

The offset- and cursor-based pagination iterators stopped as soon as a
page's `items` list came back empty. But under dataset filters (`clean`,
`skip_empty`, `skip_hidden`), the API can *scan* a full chunk (up to
1000 items) while *returning* none — `items=[]` with
`x-apify-pagination-count: 1000`. The iterators broke on that page and
silently never yielded any item behind it, even though the offset
bookkeeping (`max(count, len(items))`) had already advanced past the
scanned rows.

Now all four iterators (offset + cursor, sync + async) terminate on the
number of items *scanned* rather than *returned*: they keep advancing
the offset/cursor across fully-filtered pages and stop only when a page
scans nothing. The `limit` and `cursor is None` checks are unchanged.
Behavior is identical for every response without a divergent `count`
(collection endpoints, `ListOfRequests`).

Added a regression test for each of the four iterators.

Related to https://github.com/apify/apify-core/issues/27324.

*✍️ Drafted by Claude Code*