We know that Russian wrong-keyboard queries happen on English Wikipedia, too, and the original plan involving TextCat for language detection included enabling "Wrong-Keyboard Russian" as a language for English Wikipedia.
Instead, using second-try search we can allow the just the Cyrillic-to-Latin (RU/QWERTY) mapping on English Wikipedia (and maybe other wikis).
However, we need to create a one-way mapping in the Second Try Search library to allow Cyrillic-to-Latin mapping but not the reverse.