feat(backend): Simplify database tables with JSON-columns

[pekasen]

Situation as is: Data is stored in custom tables which generated at runtime, which is not desirable. All data bases we are targeting are able to write/read JSON-columns.

Pro:

1. This alleviates the necessity to verbosely specify columns beforehand.
2. Specifying only one table per step in the data pipeline allows faster adaptation to multilayer workflows.

Contra:

1. This may lead to unknown/uncertain columns in the data.