Run map/reduce job to populate initial index table if data table is big

Instead of internally running an UPSERT SELECT to initially populate the index table when the data table is big, we should run a map/reduce job. That way, in the event of failure, the job can pick up where it left off. Plus users will have better visibility into the % completion. Once we start collecting stats, we'll have a better idea of the size of the data table when an index is created and can make the decision based on that.