Local Sensitivity Assessment of Discrete-Time Cox Regression to Informative Censoring with Application to an HIV-Care Engagement Study