Comparison of prospective and retrospective methods for haplotype inference in case‐control studies