Issues with the measurement of informal care in social surveys: evidence from the English Longitudinal Study of Ageing