According to the commented code:

This is extremely slow and should be soon replaced for another method
that doesn't scan all Z60 objects. Like, for example, implement aliases and
include them in the secondary labels database (T262091) and add aliases to all
Z60s so that their aliases include all the possible language codes associated
to that given language (T283605)

The two mentioned tasks are completed, and yet we don't have a very efficient way to retrieve languages given their code.