Text this: Addressing identification bias in the design and analysis of cluster-randomized pragmatic trials: a case study