“Works on My Machine”: A Case Study of Replicability Challenges in Computational Humanities Research