[BUG] ISO 639 code for undetermined languages should be 'und'

[thatfloflo]

Describe the bug
We currently have zzz as the code for an undetermined language in generic task versions. According to ISO 639-2 and ISO 630-3, zzz is simply unassigned currently but could potentially be assigned in the future. ISO 649-2/3 have four specially reserved codes which will never be assigned to a specific language or collection of languages and which have defined semantics:

• mis for languages which do not yet have a code assigned, but could in future.
• mul where content is in multiple languages and a single code cannot be applied more sensibly.
• und where the language is undetermined, i.e. it is not known or hasn't been coded what the language is.
• zxx where content is non-linguistic and/or a language code is not applicable.

Of these, und clearly meets the case we have where it is not determined in advance what the other language is. We should thus adopt und in place of zzz for our 'generic' versions.

Additional context

• Library of Congress (Registration Authority for ISO 639-2) FAQ on zxx and und
• SIL entry for und
• Wikipedia on ISO 639-2/-3 generic codes