Fix broken retries in ansible tasks

[robrap]

A/C:

• All existing Ansible retries in the configuration repo use the correct syntax

Here is an example of task with retries that was broken, and PR with fix:

TASK [Tag EC2 instance] ********************************************************
...
fatal: [localhost]: FAILED! => {
    ...
    "response_metadata": {
        ...
        "retry_attempts": 0
    }
}

We might need some eSRE education around this issue.

Here are a list of other broken retries that probably need to be fixed:

$ pip install yq
$ find -name '*.yml' | while read -r path; do output=$(< $path yq '.. | select(.retries? != null and (.register == null or .until == null))'); if [[ -n "$output" ]]; then echo "$path"; fi done
./playbooks/continuous_delivery/launch_instance.yml
./playbooks/roles/mongo_3_6/tasks/main.yml
./playbooks/roles/mongo_2_6/tasks/main.yml
./playbooks/roles/mongo_4_2/tasks/main.yml
./playbooks/roles/redis/tasks/main.yml
./playbooks/roles/ad_hoc_reporting/tasks/main.yml
./playbooks/roles/mongo_4_4/tasks/main.yml
./playbooks/roles/mongo_3_2/tasks/main.yml
./playbooks/roles/mongo_4_0/tasks/main.yml
./playbooks/roles/mongo_3_4/tasks/main.yml
./playbooks/roles/mongo_client/tasks/main.yml
./playbooks/roles/python/tasks/main.yml

[rgraber]

Fixed with openedx-unsupported/configuration#7049