A repository may be in the queue to be cloned, but not yet cloning. As such
the cloneInProgress value will be false. The error checking code in the webapp
would treat cloneInProgress being false as the repo is cloned. Therefore it
treated resolveRev to uncloned repositories as a missing revision.
The only way to get back missing revision when resolveRev on the default
branch, is if the repository has no commits. So this lead to uncloned
repositories being reported as empty repositories.
This commit checks if the repository is cloned. If not, it returns a
CloneInProgressError. This seems a bit hacky, but in fact works well. Since
the page presented will update when the repository is cloning. Additionally
the repository is placed in the high priority clone queue.
Test plan: Added a large number of repositories to be cloned. Visited a non-cloned repository and observed fixed behaviour.
