I think there's a pretty huge difference between launching with two different storage capacities (Wii U) and launching with a console and a device to "make your games not look like shit". Something that significantly boosts computing power is going to look like a necessary upgrade to many, especially if they're released side-by-side, and I think that may turn people off. Getting a Wii U Basic meant "if I go digital, eventually I'll need a new external hard drive". Getting an NX at launch without an SCD means that for the time being all of your games are going to look and perform worse than they could be. It's going to directly diminish the experience, and mean people day one are going to feel forced to pay $100-$200 (or however much these will cost) to make their games perform as they should. I think Nintendo will target a single spec first, then down the line offer enhancements so it genuinely feels like an upgrade, rather than making an NX without and SCD attached feel like a gimped device built to cut costs and make you pay for the rest of the specs separately.
PC developers aren't creating 3-6 "builds" of everything, but since they're working with a platform with virtually infinite possible specs and configurations, it's a massive pain in the ass to optimize for. It's not as complex as building entirely distinct versions, but it's very much not as simple as "alright, we'll just toggle this variable down aaaaand done" - at least if you want a well optimized experience. It's part of the reason minimum specs have been ballooning on PC. Realistically, given time and resources they could get the spec requirements down, but the more hardware you're supporting, the more time and resources are going to be needed to pull off every optimization trick. It's just too much of a pain in the ass to make sure you get optimal performance from every single possible rig. Which is why a single spec hardware platform is nice. You work on optimizing everything on an extremely specific piece of hardware, meaning you can milk a lot more out of it a hell of a lot easier.
So, the reason developers are upset is because the benefit of console development is you're developing towards a single spec. Now, even if they go with the absolute minimum
approach for "NEO Mode", it's still at least a little more work since they have to test their performance on a second SKU. For instance, perhaps the game may only have certain physics bugs when it's outputting at 60FPS, etc.. Maybe "Neo Mode" lets them bump up the rendering resolution, but now they have to make sure everything behaves the same. All sorts of stuff. So no matter what, there's at least extra testing costs that are going to happen. And of course most PS4K owners are going to expect significant performance gains, and be deeply upset by anything but. So now it's expected that developers spend the extra time to further optimize for the high-spec SKU. It's just stripping a layer of simplicity from the whole thing, and under any scenario it's going to be an inconvenience.
Nintendo's daisy-chained SCD approach is much more complex, and now developers would have to worry about performance across anywhere from two to potentially dozens of possible specs. It suddenly makes a single console just as annoying to optimize for as PC. Maybe this user will have just the base console. Maybe they have 10 of the latest SCDs. You're going to have a huge gulf in performance and it's going to be expected that you optimize reasonably well for all possibilities.