I know that MUL is the suggested ISO 639-2 code for multilingual documents, but what about the 2-letter equivalent in ISO 639-1? I see in the related Wikipedia article that MU is unassigned (and thus it could be used), but I cannot find any article online on what's the recommended practice.
It is a misconception that an unassigned code in a standard can be just used for anything you want. It may be officially assigned tomorrow! This is unlikely in the case of ISO 639-1, but this is the way standards work. An unassigned code MUST not be used, and you are out of the standard immediately when you do so. An application reading the unassigned code may just stop working and throw an error. Some standards reserve some code space for "private use", ISO 639-2 has such a space for the code range qaa-qtz.
When a given standard (in this case ISO 639-1) does not fit your needs, use one that does (in this case ISO 639-2 with three letter codes).