Report "dangling" content validator errors

[DylanBaker]

Change description

We currently prune the LookML tree when building the project. This means that content validator errors relating to explores that don't exist anymore don't get reported by Spectacles. This PR changes this functionality so that these errors now appear.

The changes work in the following way:

• When we build the project for content validation runs, we now return the entire LookML tree. This allows us to attach all errors to their relevant place on the tree, even if the explore has been excluded from the run. This is what allows us to distinguish between explores that have been deleted and explores that have been excluded by the filters.
• When we attach the errors to the LookML tree, we first check if the relevant explore exists. If it does, we attach the error there. We then check if the relevant model exists. If it does, we attach the error there. This second part is new. Previously, we were only attaching to existing explores.
• When we return the errors for the project, we now allow for filters to be passed to the get_results function. This will only happen during the content validator. When this occurs, we prune the errors to just the relevant LookML tree.

This changes mean that we will now return all content validation errors relevant to a project, except for those that are linked to models that don't exist anymore. There is no way for us to know where to attach those, so they will remain dangling.

Type of change

• Bug fix (fixes an issue)
• New feature (adds functionality)

Related issues

Closes #240

Checklists

Security

• Security impact of change has been considered
• Code follows security best practices and guidelines

Code review

• Pull request has a descriptive title and context useful to a reviewer

[joshtemple]

Thanks for your work on this! I have some feedback below. Also, regarding tests: have you added any tests that explicitly test for this situation? That would be good to do in addition to refactoring existing ones.

[joshtemple]

I'm not sure how important this is, but we loop through models 4 times (441, 459, 467, 477). Do you think we can consolidate some of these loops? I don't consider it blocking for the PR but if you have some time it might give us a bit of efficiency and readability.