Uniform Inference in Linear Error-in-Variables Models: Divide-and-Conquer