Pega Robotics will only match items that are on the page, so once you leave a page, those items are unmatched. If you need to collect data from the table as you iterate through it, then you'd need to store it somewhere (a LookupTable object on the Advanced tab of the Toolbox would work perfect here). Generally, you'd iterate through the rows on your table, storing them wherever you like, and then clicking the next page link or button while waiting for the next page of results. Be careful to use the WaitForEvent for the Destroyed event for the page when navigating as you are essentially navigating back to the current page and a WaitForCreate will return true immediately before the page you are leaving has had a chance to get unmatched.
Generally, that error is benign. You can acknowledge it (it may appear multiple times) and continue.
I don't understand what you mean by this won't work. The idea is that you record the data on the current page, then navigate to the next set of records (presumably on the same page but requiring a re-load of the data) and store them as well, until you have completed extraction of the data from all of the pages.
I would break this apart into two separate automations;
An automation to control the pagination (i.e. it figures out how to get the next set of records loaded and waits until they are loaded before continuing)