I think psychologists do a disservice to their profession and their own patients when they use their credentials to diagnose people they haven't examined.
So judgement criteria can only be applied effectively to observations of in person behavior and never to observations of recorded behavior no matter how varied and voluminous?
That is saying that 4 or 5 hours of personal interaction necessarily holds more behavioral clues than tape of events, interviews, etc. that played non stop run over a calendar year.
I would say that those two scenarios are categorically different enough that the skills and criteria would have to be applied differently and that giving conclusions obtained through extensive remote observation might need language to differentiate it from conclusions arrived at through standard sessions. However the condition is with the subject and a diagnosis is recognizing a pattern of behavior. Given that a duty to warn sits as high as other ethical standards in the profession and the nature of the circumstance of the subject in the case there needs to be a pathway for psychologists to weigh in.
Abstaining does not eliminate remote diagnosis. It just leaves the field to untrained amateurs who have already demonstrated they will not hold back.