Fallback to title from browser-specs in crawl extracts

[tidoust]

Reffy extracts the title of specs it crawls. That's useful when running Reffy against a spec it knows nothing about (in other words that isn't in web-specs). Once in a while, Reffy cannot extract the title of a spec because something went wrong. Title produced in that case is "[No title found for...]". When that happens, Reffy should rather use the title from web-specs (if the spec is in web-specs, of course) in the extracts it produces.