Measurement error in environmental exposures: Statistical implications for spatial air pollution models and gene environment interaction tests