Fetch data source files from S3

[mattdelsordo]

Mapnik can use shapefiles/rasters as data sources with no issue, but these need to be stored somewhere if you want this to work in production (since the Lambda bundle size we're working with is already so large). It might be as simple as just using an S3 url as the file's location in the configuration file but this should be investigated if not implemented.

Caveat: using shapefiles has bad performance implications because of the necessity of reading the whole file every time some filter needs to be applied.