A full-site test that aims to carry out a set of tests against every URL for a site needs to get the URLs from somewhere.
Up until just a few minutes ago, the URLs to test against were collected from a site's sitemap.xml file. This was fine for sites that had sitemaps. Many don't. I'd say most don't.
For sites that have no sitemap, we collect what URLs we can from the news feed - either the RSS feed or ATOM feed.
The URLs present in a news feed are likely a subset of all the URLs for a site and are commonly a small subset.
Nevertheless, a small subset of URLs to test against is infinity percent better than no URLs to test against.
And given that HTML validation errors are common to many pages of a site, due to there being common page elements such as navigation, if you fix the errors for pages listed in your news feed you'll likely be fixing errors present on pages that weren't directly tested.