FACTS Benchmark Suite: Systematically evaluating the factuality of large language models