On the Use of General Control Samples for Genome-wide Association Studies: Genetic Matching Highlights Causal Variants