Tries to determine the current encoding based on the content. Only really helps with the UTF variants. Returns null if it can't be reasonably sure.