I don't see any particular inconsistency among what you've written and what I've written.
Certainly a labrum tear isn't 100% diagnosable, and that includes diagnosis by arthoscopic examination. However, as the severity of the tear or fraying increases, the likelihood of a correct diagnosis (a true positive) goes up. So the 80% figure for MRI exams (meaning a 20% false negative rate) has an implied threshold level of severity.
The ortho guy says 100% had labrum tears. Even if the diagnosis had perfect accuracy, there is still statistical uncertainty to consider. It the sample population is 50 pitchers, there is a 1/6 chance that the true number of pitchers with labral tears (as hypothetically measured from a much larger sample) is actually 45/50, or 90%.
As for false positives, again that depends on the observed severity. However, I think it is generally true that a false positive diagnosis happens more frequently when a patient has symptoms that need explaining. But because the team is doing MRIs on every pitcher, all of whom were actively pitching at the time, with no apparent injury, the pressure to come up with some kind of diagnosis is reduced.
That's the interesting part: Whether you think "100%" really means 90% or 80%, the majority of incoming MLB pitchers have labral tears. And they are mostly asymptomatic. So the surgeon says that labral tears are common, and the important issue is whether it causes a functional problem.
Of course, most people who have shoulder MRIs do have a functional problem. A diagnosed labral tear may be the sole cause of the problem, or a contributor, or sometimes unrelated.