split out dev/prod locations for staging and output buckets

[ohnorobo]

This separates out the DEV/PROD buckets used:

• for temp writing during the pipeline
• for writing output files from the pipeline

This does not separate the input/metadata buckets. Those are still all in the dev project.

tested: e2e test, running some python3 -m pipeline.run_beam_tables ... examples