Allow worker to write files to a bucket in prod, local filesystem in dev

[phargogh]

We should be able to have a consistent interface for where files are stored whether we're in a development environment or in production.

Possibility for helping to separate dev/prod configuration: https://pypi.org/project/python-decouple

[phargogh]

For the prod part, it appears that we can access private GCS resources from within a container on GCS with no extra work! I'm relieved. To reproduce on our VM:

$ docker run --rm -ti urban-online-workflow_worker /bin/bash
# gdalinfo /vsigs/natcap-urban-online-datasets/nlud.tif
Driver: GTiff/GeoTIFF
Files: /vsigs/natcap-urban-online-datasets/nlud.tif
       /vsigs/natcap-urban-online-datasets/nlud.tif.ovr
Size is 153811, 96525
Coordinate System is:
PROJCRS["Albers_Conical_Equal_Area",
...

writing to the bucket, however, appears to be a different task entirely and right now I'm struggling to give the worker VM's service account the permissions needed to be able to write to the bucket.

• Use cloud VM's credentials to push to a bucket.

[dcdenu4]

@davemfish, let's revisit what we'd like to accomplish with this and more generally with a dev vs production setup.