How to use the manubot.util.read_serialized_dict function in manubot

variables = {}
    for path in paths:
        logging.info(f"Reading user-provided templating variables at {path!r}")
        # Match only namespaces that are valid jinja2 variable names
        # http://jinja.pocoo.org/docs/2.10/api/#identifier-naming
        match = re.match(r"([a-zA-Z_][a-zA-Z0-9_]*)=(.+)", path)
        if match:
            namespace, path = match.groups()
            logging.info(
                f"Using the {namespace!r} namespace for template variables from {path!r}"
            )
        try:
            if match:
                obj = {namespace: read_serialized_data(path)}
            else:
                obj = read_serialized_dict(path)
        except Exception:
            logging.exception(f"Error reading template variables from {path!r}")
            continue
        assert isinstance(obj, dict)
        conflicts = variables.keys() & obj.keys()
        if conflicts:
            logging.warning(
                f"Template variables in {path!r} overwrite existing "
                "values for the following keys:\n" + "\n".join(conflicts)
            )
        variables.update(obj)
    logging.debug(
        f"Reading user-provided templating variables complete:\n"
        f"{json.dumps(variables, indent=2, ensure_ascii=False)}"
    )
    return variables

Fields in `manubot` are either generated by Manubot or hard-coded by the user if `metadata.yaml`
      includes a `manubot` dictionary.
    - All fields from a manuscript's `metadata.yaml` that are not interpreted by Manubot are
      copied to `variables`. Interpreted fields include `pandoc`, `manubot`, `title`,
      `keywords`, `authors` (formerly `author_info`, now deprecated), `lang`, and `thumbnail`.
    - User-specified fields inserted according to the `--template-variables-path` option.
      User-specified variables take highest precedence and can overwrite values for existing
      keys like `pandoc` or `manubot` (dangerous).
    """
    # Generated manuscript variables
    variables = {"pandoc": {}, "manubot": {}}

    # Read metadata which contains pandoc_yaml_metadata
    # as well as authors information.
    if args.meta_yaml_path.is_file():
        metadata = read_serialized_dict(args.meta_yaml_path)
    else:
        metadata = {}
        logging.warning(
            f"missing {args.meta_yaml_path} file with yaml_metadata_block for pandoc"
        )

    # Interpreted keys that are intended for pandoc
    move_to_pandoc = "title", "keywords", "lang"
    for key in move_to_pandoc:
        if key in metadata:
            variables["pandoc"][key] = metadata.pop(key)

    # Add date to metadata
    now = datetime_now()
    logging.info(
        f"Using {now:%Z} timezone.\n"