fix: repair kill-instance scope action slug and add instance_id parameter

[fedemaleh]

Summary

The per-instance kill action for scopes was broken. It had been renamed to kill-job-execution, a slug the API/frontend do not recognize, so the platform never injected the selected instance's context. Without a deployment_id the action routed to the scope entrypoint and failed with failed to read workflow file: .../scope/workflows/kill-job-execution.yaml.

This PR repairs the action across all scopes (k8s, scheduled_task, azure, azure-aro):

• Rename the action to the singular kill-instance — the slug the API and frontend expect to render the per-instance kill button and inject the target instance. Reverts the kill-job-execution rename (and its standalone-workflow entrypoint fallback).
• Declare the required instance_id string parameter in each action schema. The scripts already read this field (its value is the pod name, matching id from instance/list), but the schema never declared it.
• Keep scheduled_task as an override of the k8s base workflow (action: replace), so it uses its own Job→CronJob ownership check instead of the base ReplicaSet→Deployment one.

Test plan

• bats k8s/deployment/tests/kill_instance.bats scheduled_task/deployment/tests/kill_instance.bats → 16/16 pass
• bats scheduled_task/deployment/tests/workflow_overrides.bats service/deployment/tests/entrypoint.bats → pass
• All four kill-instance.json.tpl specs validated as JSON with instance_id required