The actual claim is that it is better if science is easily reproducible, but sometimes you just have to make do with what's feasible.

Well, the goal is fair: It would be nice to have science reproducible. There are simply a lot of pragmatic (and ethical, and legal ...) reasons why it often is not.