The scanner matches gtk-doc sections which match the lower case type name
to the type and uses that for the type documentation. The only problem is
it only takes the docs and none of the other annotations like deprecation
info etc.

This changes things to also parse the annotations in that case and adds some
tests while at it.