CI: Add docbook-cs linting

[jordikroon]

As announced here: https://news-web.php.net/php.doc/969388652
See for test implementation: #5510

The sniffer includes:

This also makes check-whitespaces.xml obsolete, as the sniffer will check for that as well.

The sniffer also covers #5445

• <simpara> should be used instead of <para>
• Exception or error class names wrapped in <classname> that should use <exceptionname> instead.
• Whitespace checks (.github/workflows/check-whitespace.yml became obsolete)
• Ensuring that when an element has both xml and xmlns attributes or xmlns variants, xml appears first (covers Add CI warning for xml:id before xmlns attribute order #5445)

[TimWolla]

Some remarks for the workflow definition. Didn't look at the tool itself, but the premise sounds good to me.

[TimWolla]

Some more quick GHA comments. Will need to think about the git / checkout usage more at a later point.

[jordikroon]

I am planning to merge Docbook-cs in a few days as all feedback has been addressed. Please let me know if there are comments, objections or concerns regarding this new tool. So to say any feedback is welcome.

[TimWolla]

I wanted to double-check what github.base_ref does. It is documented as:

The base_ref or target branch of the pull request in a workflow run. This property is only available when the event that triggers a workflow run is either pull_request or pull_request_target.

This means that it doesn't do the right thing for the push: trigger. Can you double-check on your fork with by removing the branch restriction on the push: trigger and observing the CI for the feature/docbook-cs branch in your fork?

[jordikroon]

I have been doing some tested and that variable is indeed empty on push.

Any thoughts on this commit before I implement it in this PR?
jordikroon@ba77f57

In short. It fallbacks to github.event.before. And github.event.pull_request.base.sha also seems a little bit better as it will just pick the hash instead of the branch name. On push it will compare against the previous commit hash. On PR it will compare against the base sha.

[TimWolla]

Can you try ${{ case(github.event_name == 'pull_request', github.event.pull_request.base.sha, github.event_name == 'push', github.event.before, 'INVALID') }} or similar for clarity? Other than that it looks good to me, I particularly add the added shell quoting (that would've been a follow-up request, depending on how your solution would've looked).