I'm not sure about the functionality of that ansible-playbook and the local-configure.yml file. Does this require merging, deployment, reload or the like?
IMO, for documentation we should have something much more agile.
Zope allows to traverse the subpath. Why not have Python methods which do so, and check for matches in a relational table?
E.g., we could have a documentation script, living in the root of https://plone.org, which of course knows about its name, and checks the table for an exact match for documentation and the subpath. If there is such a match, a redirection to the new location is performed (Plone 4 version, I'd say, because those old pages will almost certainly refer to Plone 4 or earlier). If there is no such exact match, (after a little "sorry" message) a form could be presented where the user can suggest a redirection target; administrators can view those targets and choose the best one.
Redirections which swallow path information (like /products to /download) are evil and should be removed. They are worse than nothing.
I don't know the new location, thus I can't create a pull request.
Anyway, IMO we should be able to collect broken links (and the new locations) without them; see my suggestion about the database-based solution.
If someone else were able to file issues in the docs repo that would be great! Tobias, thank you for finding these, but unless they're tracked at github.com/plone/documentation/issues your detective work will get lost in the shuffle.
Wildcard has kindly donated regular SiteImprove.com reports for plone.org and plone.com, and I'm actively fixing issues on plone.com but haven't enough time to manage the fixes for plone.org. But I see quickly above that Sven is running broken link checks on docs.plone.org already. I don't know who's actively fixing what's found in those reports.