I'm in 100% agreement.
Trying back to the Maple Leafs as an example, it's even possible (likely) that they have their own internal analytics they're using when making decisions that we're not privy to. Something like that would change everything in terms of how #3 is interpreting decisions.
I've got a few big gripes with how enthusiasts use the data and with the data we have access to.
1) It assumes that all teams are trying to do the same things within the game and it ignores any differences in approach. The simplest example is when Corsi was being used, more shots = better team. This completely ignores teams who were intentionally taking less shots, so the analytics would suggest they're bad teams.
2) They've been used as a proxy for other information they're trying to gather. I'll stick with Corsi here. It was supposed to measure possession statistics, but it was using shot attempts to do so. Shot attempts do not directly correlate with possession. Why not just actually count the seconds of possessions? Now with the technology advancements you're referring to these proxy stats are basically meaningless since there's a real time stat that gives you the information the proxy was supposed to provide.
3) I think trends in hockey are cyclical. Certain strategies work well against others and perform worse against a different strategy. As the game shifts from one style to another the analytics used by enthusiasts will always be lagging behind. What was popular today might not be tomorrow and consistently evaluating teams with the same analytics could lead you down the wrong path.
4) The predictive nature of the stats simply isn't good. I've mentioned this in conversations with other posters when it's mentioned that "it's the best we have." Flipping a coin might be the best we have but it's still not good. Dom Lucyozaodfjadjfaodf has his model that he's constantly tweaking year after year, which makes it pretty much useless because there's no solid baseline. I think trying to track a team's analytics year to year verses the entire NHL is ineffective because of roster turnover and the coaching carousel. Teams doing something unsustainable over multiple seasons isn't unsustainability, it's just that teams improve and regress over time. At some point all teams regress as well, so if you're predicting something to be unsustainable then on a long enough timeline you'll always be proven correct.
Ultimately, I think the enthusiasts are a little too bullish on their claims. They often state them as empirical truth when they're anything but. We'll see more information come from the technological advancements and teams will start to use that more than they already are (shot velocity, release times, cross ice passes, puck pathing, etc.).