* Changes to benchmarking allowed the source (repo) for a particular
  branch to be changed and as such multiple (different) versions of
  the same source reference could be benchmarked using remote and
  local repositiories (for example).

* To differentiate between different sources a unique referencing field
  was added into the results file for each result set to clearly
  differentiate.

* Since the overnight cron job retains a history of all results a mixture
  of old and new results were present and as such both had to be handled.
  This required additional logic to handle data sets that did and did not
  have unique referencing information.

* A cleanup of retained results has meant that the old datasets (prior to
  the introduction of the unique referencing) have been archived and should
  not effect newly generated data. As such the underlying logic can be removed
  from the digest or results.