The old create_colection_pages. Creates now HTML files in cache, to be read by Python search interface. No PHP anymore. Updating of collection cache and web page cache is now splitted into two, cf. usage arguments. The implementation is naive, still. But never mind.