docs: add async response_mode to workflow API reference

[Umair706]

Summary

Add documentation for the new response_mode: "async" option in the workflow execution API, introduced in langgenius/dify#35301.

Changes to en/api-reference/openapi_workflow.json

• Add "async" to the response_mode enum in WorkflowExecutionRequest schema
• Update response_mode field description to explain async behavior
• Add request example for async mode
• Add 202 Accepted response to both /workflows/run and /workflows/{workflow_id}/run
• Add WorkflowAsyncResponse component schema (returns task_id, workflow_run_id, and data with status: "scheduled")
• Update 200 response descriptions to cover idempotency cache hits (duplicate async requests return 200 instead of 202)

How async mode works

1. Client sends POST /v1/workflows/run with "response_mode": "async"
2. Server returns 202 Accepted immediately with a workflow_run_id
3. Client polls GET /v1/workflows/run/{workflow_run_id} for results
4. Status progresses: scheduled → running → succeeded / failed
5. Optional Idempotency-Key header prevents duplicate executions on retry

Test plan

• Verify openapi_workflow.json is valid JSON
• Verify Mintlify renders the new enum value, 202 response, and schema correctly