A cautionary note on ignoring polygenic background when mapping quantitative trait loci via recombinant congenic strains