Rasters and saved and updloaded to s3 in a loop now, rather than each…

… one being hard-coded. Still trying to figure out the large unmanaged memory issue that's making this crash on large areas or many years.