Speed up used languages calculation on charts page
What does this MR do?
Fix our last user of Rugged::Repository#fetch_attributes: https://performance.gitlab.net/dashboard/db/method-calls-by-action-and-worker?orgId=1&var-method=Rugged::Repository%23fetch_attributes
Are there points in the code the reviewer needs to double check?
Well, do we want to monkey-patch this? We should probably also make this chart load async, and cache the chart data, but this is an easier change
Why was this MR needed?
https://gitlab.com/gitlab-org/gitlab_git/commit/340e111e040ae847b614d35b4d3173ec48329015 wasn't being used in one place, from a library: https://github.com/github/linguist/blob/v4.7.0/lib/linguist/lazy_blob.rb#L33-L36 I think we can be reasonably confident that this will still produce correct results, and even if they differ slightly, we don't document these special Linguist attributes