Description

Currently, if the dbt project contains seeds, Cosmos will render them and attempt to run dbt seed unless the end-user customises how Seeds should be rendered:
https://astronomer.github.io/astronomer-cosmos/configuration/render-config.html#customizing-how-nodes-are-rendered-experimental

However, in most cases, seeds do not need to be continuously run.

It would be great if Cosmos could allow users to opt for different ways of rendering/running seeds, similar to source nodes:

• SeedRenderingBehavior.ALWAYS: Current behaviour of Cosmos, always add the seed tasks in the DAG/TaskGroup and run seeds
• SeedRenderingBehavior.NONE: New behavior, don't render any seeds in DAG/TaskGroup
• SeedRenderingBehavior.WHEN_SEED_CHANGES: Only render and run a seed if the csv file has changed since the last execution

There are a few open questions:

• Should we still render seeds so users understand the topology, even if we don't run them in SeedRenderingBehavior.NONE and SeedRenderingBehavior.WHEN_SEED_CHANGES, and use an EmptyOperator in those cases?
• Where could we store the state if a .csv was run and if it changed? As an Airflow variable? In a remote object store? Somewherelse..?
• It would be probably neat to render the source nodes, but skip them, unless they should run

I'd love your thoughts on these ideas.

Use case/motivation

Allow users to opt out of running seeds and only run seeds when needed.

Related issues

No response

Are you willing to submit a PR?

• Yes, I am willing to submit a PR!