Now when our code is all UTF-8, it would be good if we provided hints across all the jobs that they should use UTF-8. The argument here is that providing that hint is better than leaving locale selection undefined. My suggestion is to set LANG=en_US.UTF-8 as a global env var. This should provide a default for all the LC_* ones if not set. Another alternative is to set LC_ALL=, but that would override all LC_* even if set by the site, and we might want to reserve that for users to set.

This obviously also needs to be documented.