The workflowId is stored in the task_identifier. It happened by chance that the task_identifier matched the node_id in the subworkflow used for the test:
When we drop a new subworkflow as a node in a graph, generally, the node_id doesn't match the task_identifier: