Cloning the full (partition) repository takes a significant amount of time. This is only blocking on network reads and we can leverage the CPU resources by running tests while continuing the reads. The approach is to use a sparse checkout and checkout seed directories while tests are running on those which have already been checked out.