Add subcategory functionality to PNI categories.

[Pomax]

Closes #7617

Proper admin interface sorting will be tackled in #7640

Note: this can't be testing in the review app, because the only way to verify the values work for templating purposes is to already have templates written, which we can't because that depends on this existing. So... local testing!

Testing:

• create a bunch of PNI category snippets, with the "parent category" set to something.
• add this into the catalog.html template so you can actually see these values doing what they should be doing:

<h1>children for {{ current_category.name }}:</h1>
<ul>
  {% for child in current_category.get_children %}
    <li>{{ child.name }}</li>
  {% endfor %}
</ul>

• And then navigate to the category page that you made some test subcategories for. All of those should show up.

Also note that in the admin interface, PNI snippets should be ordered as "parent categories first, alphabetically ordered", followed by "and then all sub categories, with a reverse alphabetical parent order, and regular alphabetical ordering within that. This is because of the Meta class rule ordering = ['sort_order', '-parent__name', 'name', ] where the parent__name needs to be reverse because if there is no parent, that parent__name is None and None comes first, which would be super not-useful.

Testing part 2:

• you should be able to make new categories without setting a parent category.
• you should be able to assign both toplevel categories and subcategories to products.

[richbrennan]

Works as expected, thanks @Pomax !

A couple of observations that might be considerations for later:

• I can create arbitrarily-deep trees of categories
• I can set a child-category to be a parent of its parent, creating a circular hierarchy

In case it might be useful, I've been meaning to investigate this for nested categories: https://posts-by.lb.ee/building-a-configurable-taxonomy-in-wagtail-django-94ca1080fb28

[Pomax]

@richbrennan I think in this case for (1) we're okay with that, and for (2) let's land this first, and then make a snippet save reject the data if that introduces a cycle in follow-up. Given that we have a fixed ontology that we're going to implement with this mechanism (see link in issue) we shouldn't be running into cycles yet, giving us plenty of time to hardned the snippet's form handling/save logic.

I've filed #7644 as followup for that work.

[danielfmiranda]

Hi Pomax! Thanks for the detailed instructions, I followed them and tested everything out, and can confirm that everything is working as expected! Approving 👍