Risk-sensitive stopping problems for continuous-time Markov chains