standardise passing down config parameters

[galamit86]

Decide on and implement "best-way" to pass down configuration information - at this point that includes paths and 3 nested GCP environments.

Current thinking is hand down full object config, and possible extra parameters if needed - currently gcp_env is required to choose between the 3 different GCP possibilities (dev, test and prod).

Another option is to create paths and gcp and hand only the required pieces to the relevant function-tasks

---

Extra info about Config:

Currently, the config parameters are inputted via nl_open_data/user_config.toml, and added to Prefect's global config object at import (via config.py or __init__.py - TBD). Within a certain flow-script, i.e. regionaal.py, the config object is imported:

from nl_open_data.config import config

[dkapitan]

@galamit86

I am not quite sure I follow your logic to overwrite the imported from nl_open_data import config object with a new Box object?

+1 on just passing the whole config object to each flow and task
+1 to add [dev|test|prod] as a parameter

[galamit86]

You're absolutely right, that's not needed, bad logic :)

[galamit86]

• Use Box and toml
• Remove dataclasses altogether, removing serde and tomlkit along the way