But both of those cases are concerning the “Bayesian epistemology” side of things, not using probability to reason about how to bet in games of chance with well

But both of those cases are concerning the “Bayesian epistemology” side of things, not using probability to reason about how to bet in games of chance with well-established rules and mechanics. The usage of probability Deutsch is approving of is the latter, and is unrelated to a particular observer’s level of confidence that the cards are designed fairly or that the dealer and other players are playing fairly.