I have an inkling of an idea of how a true-ish 3DS successor might not only work, but also work in such a way as to both contribute to Nintendo's agenda and allow them to keep making games for "essentially one platform".
You'd have to look a little ways back at a little device called Nvidia Shield Portable.
It wasn't a particularly successful device, for a variety of reasons, but primarily I think it was because it wasn't a Nintendo handheld. You could play a lot of old Nintendo games on it, but there just wasn't the kind of first-party support for it that you would really be tempted to buy it over just a gamepad for your smartphone.
Fast-forward to this generation. The Tegra 4 is still a massive improvement over the 3DS hardware. It is, still, a Tegra device, with everything that entails. And nowadays, it is likely to be just the kind of 'dirt cheap and readily available' that Nintendo could make use of for a cheap handheld console.
It would, in a lot of ways, be a "Switch Lite", but with none of the Switch branding. It's be named something silly, like DSNeXT. But under the silly (yet memorable) naming and simplified, dedicated-handheld design, would be the same kind of backend power, OS and APIs that run on the Switch. Just, adjusted downwards to a 480p screen and some 50 GFLOPS of compute power (or however much the T4 or T4i can put out). And there'd be a full suite of controls on the gamepad/body, slimmed down and refined to better fit the handheld's purpose.
The practical upshot of such a device, would be that any existing Switch game that does not need a tremendous amount of processing power - a lot of pixel indies, VNs, 3DS-like games and re-releases, "Nintendo Classics" up to their N64 catalog if that ends up happening, all would end up straightforwardly ported to that device. And any game made specifically for that device's audience, the little mobile games, 2D platformers, continued series from the 3DS and DS - all would have an open bridge to be directly playable on the Switch, distributed via the eShop, with little to no modification necessary.
It'd be an ideal "gateway console", far more directly 'preparing' a young audience for owning their own Switch later on, with a reduced expense up-front and little to no migration cost.