If you load that URL and look at the HTML, you won’t find any h1 element within another element of class current-version, hence the unknown and the error.
Nothing wrong with the integration: the page layout has changed which is something it’s always susceptible to, and why scrape should only be used as a last resort.
The documentation should probably be updated, either to something that does work, or to make it clear that the examples are not guaranteed to return any results.
outside what troon is saying, there’s no reason to even make that sensor because home assistant provides a rest endpoint for the current versions. Lastly there’s an attribute on the update entity that also has the current version.