allowing "unused"/hidden properties in playbook schemas to facilitate more useful YAML anchors

[E314c]

The use case is this: I have some playbook which has a few attributes common between a number of steps, but in themselves are not a "complete" step. I would like to have one declaration of these attributes and reference them elsewhere in the playbook:

---
- name: My Play
  hosts: someHosts

  # This snippet is not a complete task definition, but is used to reference multiple attributes in a repeatable way
  .become_app_user: &become_app_user
    become: yes
    become_user: '{{ app_user }}'

  tasks: 
    - name: Run task as main ansible user
      ansible.builtin.shell: whoami
    - name: run task as the app_user
      << : *become_app_user
      ansible.builtin.shell: 'echo "$USER";'
    - name: perform some action as app_user
      << : *become_app_user
      some_action: to_perform
    - name: perform action as main ansible user again
      some_action: to_perform

I propose we allow attributes prefaced with . at the top level of a play, where ansible will ignore them, but they can be used for anchors and attribute merging

There was a previous request for this on the main ansible repo, but there the user was unaware of a simpler way of obtaining the same results as they were just trying to reference configuration multiple times in a host file.

I recognise the above example could be "resolved" by splitting into separate play files, but it is meant as a demonstration of a useful "power" that seems to be arbitrarily denied to users at this time: If we allow . prefixed attributes at the top level of a play, we allow for snippets to be reused.

The change to the playbook schema is small:

{
  "$def": {
    "plays" {
        "(..." : "existing schema ...)"
        "patternProperties" : {
          "^\\.": {}
        },
    }
  }
}

Is there any technical risk for this kind of thing? Would it also need a change in the ansible parser? or would it safely ignore the unused properties