XPath "in-use" language retrieval issues on unpublished nodes

[lssweatherhead]

• Fix for Variable $ancestorOrSelf not working on unpublished pages  #79 - Variable $ancestorOrSelf not working on unpublished pages Variable $ancestorOrSelf not working on unpublished pages  #79
• Adds checks for unpublished child and root content
• In the case of a non-root node being edited in an unpublished state, the code defaults to the parent node to use in the XPath (as per the existing functionality for adding a new node)
• Falls back to "all installed" language choice where root nodes are being added (as "in use" languages will not have been defined yet)

[mattbrailsford]

Looks good but just one item I would suggest changing. If you make the change, I'll happily merge it in. Thanks for the PR

[mattbrailsford]

I'd be tempted to refactor out the use of the ContentService here as this will be performing a database hit. You could probably go one of two approaches. 1) given you are really only checking to see if the content is actually published, you could check the published content cache and if it's not their, assume it's unpublished or 2) perform a lucene query on the internal lucene index which contains unpublished nodes and check there. I reckon option 1 would be easier but should do the trick I think.

[lssweatherhead]

Great, thanks for taking a look Matt - I'll update and re-commit just now :)