How to use the meteomatics.rounding.get_num_decimal_places function in meteomatics

for _ in range(num_payloads_per_forecast):
            for lat in lats:
                values = binary_reader.get(value_data_type, num_lon)
                dict_data[lat] = values

        df = pd.DataFrame.from_items(dict_data.items(), orient="index", columns=lons)
        df = df.replace(NA_VALUES, float('NaN'))
        df = df.sort_index(ascending=False)

        df.index.name = 'lat'
        df.columns.name = 'lon'

        if parameter_grid is not None and parameter_grid.endswith(":sql"):
            df = df.apply(parse_date_num, axis='index')
        else:
            df = df.round(rounding.get_num_decimal_places(parameter_grid))

        if num_forecasts == 1:
            return df
        else:
            dates_dict[dt.datetime.utcfromtimestamp(forecast_date_ux)] = df.copy()

    return dates_dict