Test link expansion inclusion rules with truth tables (direct links)

[nacnudus]

Note: this only covers direct links. We'll cover reverse links separately.

This exhaustively tests the simplest case of link expansion, when there is a single candidate link to be included in a Content Store or GraphQL response. The simplicity makes it possible to test all possible combinations of properties of link sources and targets.

This differs from other tests because it specifically checks whether a link should be included, whereas other tests only check whether GraphQL does the same thing as classic link expanion, whether that's correct or not.

Drafts are currently only tested for classic link expansion. Once GraphQL supports drafts, a minor change to these tests will test GraphQL's link expansion too.

The more complex scenario, not addressed here, is when there are several candidate links. Precedence determines which of the candidates are expanded. But for a link to be considered for precedence, it must already pass the inclusion rules tested in this PR.

[mike3985]

This doesn't look like it could've been simple to figure out 😓 Nice work.

I've left one comment, I think it'd be a worthwhile improvement.

Other than that, I do think there is room for improvement in readability / an opportunity for some of this code to be more self-documenting. But I'm happy to wait and see how it is to work with and to make improvements later, once more of the team have come into contact with it 👍

[mike3985]

It's considered idiomatic in Ruby to stick a trailing question mark on the end of boolean-returning predicate methods (I'd include with_drafts, included, etc in this group). That's going to be less convenient than declaring an attr_reader, but I think it's worth it.

def with_drafts?
  @with_drafts
end

There're probably one or two shorter-hand ways of achieving the same thing 🤔 I'd be equally happy with the approach above, though.

I'd go a step further than just the public predicate methods, too, and do the same for withdrawal and permitted_unpublished_link_type, because I think meeting our expectations for this particular naming convention will make this easier to read.