Tortoise.init_models does not raise ValueError on wrong input

Describe the bug

I was using the Tortoise.init_models() module to initialize the orm, but it was not clear to me that the first parameter had to be a list, the doc says: "Models paths to initialise". To me this could be either a list of strings or just one string if only one module is used, which is what I did, I put the string "models" instead of passing the list ["models"] as first argument.

Unfortunately strings are iterable too, and I got stuck with error messages like "module m" not found and the likes, which was not clear to me that it was just the first letter of the sting I was passing as the argument.

From the source code (v 0.17.7) the method Tortoise.init_models is expecting basically a list of strings where each string represents the import statment for each module, but any iterable is acceptable, leading to inconsistent and unexpected result.
`
@classmethod
def init_models(
cls,
models_paths: Iterable[Union[ModuleType, str]],
app_label: str,
_init_relations: bool = True
) -> None:

    Early initialisation of Tortoise ORM Models.

    Initialise the relationships between Models.
    This does not initialise any database connection.
    :param models_paths: Models paths to initialise
    :param app_label: The app label, e.g. 'models'
    :param _init_relations: Whether to init relations or not

    :raises ConfigurationError: If models are invalid.

`

To Reproduce

just try:
Tortoise.init_models("models", app_label="models")
instead of:
Tortoise.init_models(["models"], app_label="models")

Expected behavior
I would expect a ValueError Meassage saying that a string was passed as argument, while a list was expected.
Plus, I think imho that the models_paths should be declared as a List, not as a generic Iterable.

Thank you very much for the effort

Start at the Tortoise.init_models entry point and reproduce the string and list calls shown in the issue. Verify that a string produces a clear ValueError while the list form continues to initialize models.