Variability and repeatability of house dust mite allergen measurement: implications for study design and interpretation