feat: adding a json schema command

[henryiii]

Started on a schema command that produces a JSONSchema. Includes a --strict mode that doesn't allow extra unknown commands. Mostly respects and supports plugins (tox-gh and tox-uv tested locally).

• ran the linter to address style issues (tox -e fix)
• wrote descriptive pull request text
• ensured there are test(s) validating the fix
• added news fragment in docs/changelog folder
• updated/extended the documentation

Making a draft PR to store progress.

Current schema can be seen rendered into the PR at the moment.

[henryiii]

Adding a commit that would support plugins setup like tox-gh. To actually fix that one, though, it needs to be modified a bit something like this:

    bail_reason = None
    if state.conf.options.command == "schema":
        pass
    elif not core_conf["is_on_gh_action"]:
        bail_reason = "tox is not running in GitHub Actions"
    elif getattr(getattr(state.conf.options, "env", None), "is_default_list", False) is False:
        bail_reason = f"envlist is explicitly given via {'TOXENV' if os.environ.get('TOXENV') else '-e flag'}"
    else:
        logging.warning("running tox-gh")

[ssbarnea]

I would want to ensure that the generated schema is valid, you can make use of https://python-jsonschema.readthedocs.io/en/latest/validate/ for validating it. We already do this in ansible-lint with ~20 schemas if you want to take a look.

[henryiii]

I've been using uvx check-jsonschema --schemafile src/tox/tox.schema.json tox.toml to check it locally so I just need to look up a Python API for it, I'll check ansible-lint's usage. The biggest question, though; does it make sense to keep a (continually) rendered version in the repo? Or should it only be generated?