Orquesta - ChatOps - Duplicate chatops.post_result tasks triggered

SUMMARY

Orquesta worflow execute post_message twice when one task is failed. Possibly related to https://github.com/StackStorm/st2/issues/4362

STACKSTORM VERSION

st2 3.3.0, on Python 3.6.12

OS, environment, install method

Ubuntu 20.04.2 LTS
Custom install from official repo:
deb https://packagecloud.io/StackStorm/stable/ubuntu/ bionic main

Steps to reproduce the problem

Simple workflow to reproduce problem:

---
version: 1.0

description: Update AWS Parameter Store

vars:
  - task_message: null
  - stdout: null
  - stderr: null

output:
  - task_message: <% ctx().task_message %>
  - stdout: <% ctx().stdout %>
  - stderr: <% ctx().stderr %>

tasks:
  main:
    action: core.noop
    next:
      - when: <% succeeded() %>
        publish:
          - task_message: "main ok"
        do:
          - test
          - notify_slack
      - when: <% failed() %>
        publish:
          - task_message: "main failed"
          - stderr: <% result().stderr %>
        do:
          - notify_slack

  test:
    action: core.noop
    next:
      - when: <% succeeded() %>
        publish:
          - task_message: "test ok"
        do:
          - test2
          - notify_slack
      - when: <% failed() %>
        publish:
          - task_message: "test failed"
          - stderr: <% result().stderr %>
        do:
          - notify_slack

  test2:
    action: core.local cmd="exit 0"
    next:
      - when: <% succeeded() %>
        publish:
          - task_message: "test2 ok"
          - stdout: <% result().stdout %>
        do:
          - test3
          - notify_slack
      - when: <% failed() %>
        publish:
          - task_message: "test2 failed"
          - stderr: <% result().stderr %>
        do:
          - fail
          - notify_slack

  test3:
    action: core.local cmd="exit 1"
    next:
      - when: <% succeeded() %>
        publish:
          - task_message: "test3 ok"
          - stdout: <% result().stdout %>
        do:
          - notify_slack
      - when: <% failed() %>
        publish:
          - task_message: "test3 failed"
          - stderr: <% result().stderr %>
        do:
          - fail
          - notify_slack

  notify_slack:
    action: core.local cmd="echo 123 >> /tmp/debug.log"
    next:
      - when: <% succeeded() %>
        publish:
          - task_message: <% ctx('task_message') %>
          - stdout: <% ctx('stdout') %>
          - stderr: <% ctx('stderr') %>

Expected Results

Have only one result in slack with execution success/failure

Actual Results

Have 2 messages in slack.
It doesn't depend on -fail parameters count. You can fail all tasks or only one - you'll get exactly two chatops.post_result executions: https://prnt.sc/1069gou
I check both execution details, and there's no difference except end_timestamp, occurrence_time and id fields

Use the supplied Orquesta workflow as the reproduction case, then inspect how its failed-task transitions trigger chatops.post_result. Compare the two execution details, especially their timing and identifiers, and verify the workflow produces only one Slack result after the fix. No source file or test path is named in the issue.