Add another specialization for run: job or k8s-job

[cvgaviao]

What would you like to be added:

Currently, the spec includes run:container, which I inferred requires a Docker or Podman client installed on the host where the orchestration application is running.

But there is another scenario: The image may be run as a Kubernetes/Openshift Job in a cluster.

Why is this needed:
Many companies uses ephemeral Kubernetes Jobs to do batch processing.

[fjtirado]

Im not sure if this one belongs here or should be opened on the specification repo here https://github.com/serverlessworkflow/specification/tree/main
You are right that container is asuming docker or podman is installed on the host running the orchestration application and I think we need some way on the specification in order to make the distinction you are looking for when the workflow is written

[fjtirado]

My bad, you did not ask for a different idiom in the specification, but a different runner for container (we only have docker right now), certainly we can add a new one.

[ricardozanini]

@cvgaviao we should use the same run: container. Per spec, we can use it to run the jobs, pods, and ephemeral containers based on the execution context.

A specific run: job though, would require a new handler in the specification as @fjtirado commented, which I think is unnecessary.