Should raise an Exception when tokenizer is not defined

In bertopic/representation/_utils.py, line 57, tokenizer is possibly None. In this case, an exception asking the user to explicitly set tokenizer should be raised.

    if doc_length is not None:
        if tokenizer == "char":
            truncated_document = document[:doc_length]
        elif tokenizer == "whitespace":
            truncated_document = " ".join(document.split()[:doc_length])
        elif tokenizer == "vectorizer":
            tokenizer = topic_model.vectorizer_model.build_tokenizer()
            truncated_document = " ".join(tokenizer(document)[:doc_length])
        elif hasattr(tokenizer, 'encode') and hasattr(tokenizer, 'decode'):
            encoded_document = tokenizer.encode(document)
            truncated_document = tokenizer.decode(encoded_document[:doc_length])
        return truncated_document
    return document

Open bertopic/representation/_utils.py and inspect line 57, especially the doc-length path when tokenizer is None. Add the requested exception requiring the user to explicitly set tokenizer, then verify that undefined-tokenizer input raises it while the listed tokenizer modes retain their behavior.