This is true, those players wouldn't be as good individually if they were on teams with worse team defense. Matchups matter too, as does quality of opponent as do lots of other splits. Smart teams will use all these data points together, and not raw numbers.
Harden plays the worst guy in his size range from the opposing team. Any rating that has him above the bottom 20% or so of starting defenders is simply wrong.
This argument and your hatred of WAR are my two favorites. They're metrics of limited value that most analytics nerds don't particularly like, but they have some value, and pretty decent predictive value too! Doesn't mean they should be used in a vacuum or that they're even all that great. The Grizz do have a solid analytics dept, but I also think they have dealt with way too much staff turnover and awful contracts for a while, in the West, and the lottery hasn't been great to them. If I had to guess, the "guy who invented PER" has probably invented lots and lots of metrics the Grizzlies use.
This is one of my favorite arguments. PER is quoted as being nearly biblical. Hollinger was hired because of his analytics abilities and the team has regressed since his arrival.
I don't know enough about the first point except to say that there's a selection bias in all sports for people willing to take internships or low-paying analytics jobs to try and work their way up that skews rich and white. We obviously agree on the analytics being just one part of running a successful organization. Relying on it too heavily is bad! Didn't see this point in your long and hard to read post.
I wouldn't hire Jerry West to run my stats team and I wouldn't hire
this guy, who the Rockets poached from Google, to run my scouting operation.