fix: staging db_import is using prod url

[daniel-ji]

Currently, we use a prod url for https_prefix for both staging ingestion and db import. This created the issue of run, dataset, and deposition key photos using prod url prefixes, which would break before datasets were surfaced to prod.

We did this originally because we wanted our generated s3 files (the neuroglancer config being the primary case) to contain only prod url references, and we would make staging work by replacing these prod urls on the fly. We also changed the db import to also default use prod urls, which is a mistake I made in #545. For db_import, we should still default to a staging https_prefix, since the scenario of generated s3 files doesn't occur for db import, because this is just db import and not ingestion / generate files.

As a solution, I think we should go from our current default behavior:

• staging ingestion: prod url for https_prefix
• staging db import: prod url for https_prefix
• prod db import: prod url for https_prefix

New default behavior:

• staging ingestion: prod url for https_prefix
• staging db import: staging url for https_prefix
• prod db import: prod url for https_prefix

Note that currently the following datasets are affected (using prod urls for preview of the run, dataset key photos on staging): 10444, 10457, 10459-10469

The fixes in this PR was validated to be working by re-db-importing DS-10458, which it now has the correct urls (DS-10458 previously was also an affected dataset but no longer).

[uermel]

Thank you!