This fixes a bustination of the official ICQ client in at least some locales. For away (and possibly other) messages, apparently the official ICQ (5.1?) client of some locales converts messages which are stored in UTF-8 from a locale-native character set to UCS-2BE; this results in something which, when decoded "correctly", is gibberish. Instead, we first try decoding from UCS-2BE to the locale-specific character set, and if that validates as UTF-8, we display it, instead. Since UTF-8 is relatively picky, hopefully this won't break too many sane clients.