Fix collection test flake due to successful canceled command

[AlanCoding]

SUMMARY

This took some frustration to get to the root of.

- name: Wait for the command to exit on cancel
  ad_hoc_command_wait:
    command_id: "{{ command.id }}"
  register: wait_results
  ignore_errors: true

- assert:
    that:
      - wait_results is successful
      - 'wait_results.status == "successful"'

Think about that for a second. The job (ad hoc command) is canceled, and it's status is successful... how did this ever pass?

It appears to be a race condition where our cancel logic is so slow that the command completed before the cancel took effect. If it ever too too long to dispatch the command (or something like that) the test would fail. Looks extremely hardware-specific.

ISSUE TYPE

• Bug, Docs Fix or other nominal change

COMPONENT NAME

• API