The sceptical Bayes factor for the assessment of replication success