I wonder if the original export code that lead to this investigation was actually correct? It sounds like sprintf() was being called without an explicit locale. This can be fine if Blender does a top-level setlocale(), but can also be subtly and horrible unfine otherwise...

Sounds like there's plenty of opportunity for library-level improvements here (as well as application-level workarounds), but certainly the sprintf_l(..., locale, ...) being slow is the most surprising to me, and likely the easiest to fix.