On my point of disagreement, I've also seen trainers and behaviorists do the same thing. As for a different environment possibly changing the behavior, of course that's right, but it's also additional information. If the behavior is not exhibited in another environment, I would begin teaching a replacement behavior under those conditions, before later transferring it to the target environment. It is far easier to teach a new behavior without conflict, and I feel that sequence is far more time efficient. I agree, of course, with use of footage.
I recently dealt with a dog who had bitten people on several occasions and nobody knew what to do or what precautions to take. This was not a question of how far the behavior could escalate (which was somewhat obvious). The question here was why and what could be done. It took a series of specific tests to determine the issue, and the dog was then transferred to a facility having the knowledge and resources to help him. Yes, the dog was provoked, multiple times. No, he was not unduly distressed, and recovered each time within seconds. During the tests, he attacked people (me), places and things, including a building. From those results I could give them appropriate handling precautions.
As to their being plenty to observe before that point, that is not always true. In my example case, what had people confused was that the dog would be completely fine and calm one second, then lunging for a bite the next one. While not very common, this does occur. If there should instead be a great deal to observe, then I'm just fine with that.
With the more common cases, there are still the questions of intensity and focus. In recent play groups, for instance, a dog who ramps up slowly and can be easily distracted is handled quite differently from one who attacks abruptly and will ignore common adversives during her attack. In previous groups I handled with more difficult dogs, it was important to make this distinction before the dog was brought into a group setting. Unless you have this knowledge, you cannot make this distinction.
Having said all this, I will still agree that your advice and comments are valid for the greater majority of cases. Also, that I tend to work more with the minority cases. With that I submit that the degree of examination and testing required depends on the problem details, and no simple formula or single approach can cover all cases. And that most failures I have seen have been due to that limitation in their approach, and that this is fostered by many books and articles that seem to imply a particular approach is the best in all cases.