Double-calibration estimators accounting for under-coverage and nonresponse in socio-economic surveys