move to non-spark etl glue jobs when we don't use spark

[thecaffiend]

This is super low priority as the costs seem to be exactly the same, but we care currently setup to use python spark jobs for our ETL glue jobs, but we do not make use of spark. The HAI specific jobs just be plain python jobs. (those are the only etl jobs we have right now, but by time of implementation there may be more to check out)