I don't usually break up quotes like I am about to, but you have a large number of interesting points to respond to.
I'm just talking about resolution, so it's a fairly narrow domain the player is driving. The DM narrates along, only if a player character does something is there resolution - which may still be nothing more than the DM narrating the results of that action, with no reference to a check or other role.
Sorry if that's not a very deep or useful insight.
It is and isn't at the same time. The big issue that comes up frequently is when a conflict arises between multiple PCs, whatever the reason may be or how large it may be. 90% of what I impose as a DM over PCs via NPCs is usually because one stubborn player doesn't want to go along with the rest of the party, and as they argue, I lose the attention of most of the rest of the players. Having a particularly notable NPC and basically force resolution is often the only in game solution I have to keep the game moving along. Even then, I rarely force a specific resolution, as long as some kind of resolution is met, and usually just the threat is enough to get that resolution needed to keep the game moving. I find that there has to be the option of an NPC overriding player character and/or party actions, but if everything else is done right, that option is rarely called upon.
I'd expect a 5e DM to narrate NPC actions without resorting to dice as often as with PCs, since the DM does have that extra level of information, he can resort to an uncertain outcome less often.
This is where I expect the DM to separate NPC knowledge from DM knowledge just like I expect a player to separate character knowledge from player knowledge. No one NPC is going to have the full knowledge of the DM anymore than any given character is going to have the knowledge of an experienced player. Just like I expect players to work from the general position of their current character, I run individual NPCs from their respective backgrounds and knowledge bases, meaning that uncertain outcomes remain at more or less the same probability with any given NPC. No single NPC can influence the shape of the story any more or less than any single PC. A group of NPCs has no more or less influence than the full party of PCs. My full knowledge as a DM almost never comes into play until the party starts dealing directly with the ultimate boss or his direct henchmen, and by that time, the players should know almost as much as I do about the general direction of the campaign, and have ways of influencing it that the boss doesn't automatically know about. A DM that doesn't separate knowledge this way, I completely agree with your point, but that is not how I, or many, run it.
I think a good enough DM is sufficient, even working with inexperienced or casual players.
DM Empowering is not exactly the same as DM-centric. The center of attention and action is always the PCs. 5e Empowers the DM, very clearly makes the system itself sub-ordinate, building DM-dependence into its sub-systems, which as you rightly point out, is fantastic with a good enough DM, but risky until the DM gets there.
That resolution always relies on the DM to make a pass/fail/check ruling for any player-declared action is a foundational element of that.
You have had far better experience in that arena than I have if you believe that. "Good enough" in terms of both DM and player is difficult if the only time the people see each other is at the game and different people have experience at different tables. The time in the game itself is often not enough to fully get everyone on the same page. A good enough DM and all brand new or casual players can work simply because the DM isn't fighting any particular expectations that the players may have. But mix a good enough DM with players who already have experienced the game at another table, and good enough often is not enough to keep the campaign on track. My own current campaign is a shining example of this. I am a decent, but far from great, DM. I am in a group that rotates DM duties, and most of the other DMs and players know each other quite well, playing together elsewhere; I am very much one of the outsiders of the group. My campaign almost didn't get past the first encounter of getting the party together because, no matter how much I tried to tell them I did not run the same style of game that the usual DMs did, they still brought all of their normal expectations to the table. I was able to get through it because this was a 3.5 game, and the rules carried us long enough for a common ground to be found. With 5E, it would have died immediately, and it probably never would have even been attempted in the first place, because I knew that I would be fighting a lot of built in expectations no matter what I did, far more than my current skill as a DM could easily bridge in the critical first couple of sessions. Hence why I argue that "good enough" is difficult enough to pull off in 5E to the point that it basically doesn't happen. A patient pick up group, which the group above basically is, or a group of good friends will reach there after two or three false starts (this is the fourth attempt at a 3.5 based game in this group I know of, and only one other that I have been there for has gotten past the first month's worth of sessions), but most groups, especially ones with newer players, don't have that kind of patience. It's usually hook them or lose them in the period of one or two sessions. Not a lot of time for a "good enough" DM to work with.