Causal Interference in Panel Data with Application to Estimating Race-of-Interviewer Effects in the General Social Survey