Considering the design life cycle there is no doubt that Sony planned the a7rIII long before Nikon released the D850.
Odds are it is simply the result of their strategy for the a7 series of camera that will remain their mid-range as opposed to the higher end a9r.
There is therefore no reason to think you’ll be able to buy a higher res body any time soon without paying the 5,000 US$ the a9r is likely to cost. 
Cheers,
Bernard
I really don't care about the cost - more that the product is available. In any case, starting from Sony's underdog position with regards to market share vis-a-vis Canon, there's no way they could make it cost more than the 5Ds replacement and have a commercially-viable product (5Dsr's launch price was USD3900), so that more-or-less sets an upper bound on price.
The design life cycle consists of a number of different units and many different prototypes, most of which can be made ready for production with a few months' notice. It's a continuous process of refinement, improvement and experimentation, with many different prototypes in circulation at any one time, not a case of 'We'll start designing the A7r4 now, and it'll be ready in 24 month's time'. No doubt Sony would have had a number of different A7- and A9-type bodies, all with different sensors in them, to test different aspects of performance. When the time comes, they just pick the most suitable prototype, or most suitable parts from various prototypes ('most suitable' for any number of reasons), decide what to call it and move it towards production. It's the same way most hardware R&D works - the engineering team continually puts out the patents and prototypes, most of which never go to market, then, when management needs a new product, they go to the R&D team's library of prototypes (or prototype components, in the case of modular components) and take the one most suited to the role and put it into production ('most suited' whether in terms of performance, manufacturing cost, manufacturing capacity or just about any other criteria).
I suspect the D850 caught Sony off-guard - not so much with regards to its sensor, but with regards to its overall performance. They had to get something to market relatively quickly. A suitable body and suitable off-sensor electronics were ready, but the next-generation sensor was most likely not (most likely with regards to production methods or production capacity, rather than sensor design). So they just picked the most suitable prototype they could get ready for mass production in the limited time available, while the R&D team continued working on refining sensors, electronics and manufacturing capabilities for future releases. Much easier to use a sensor they're already producing, and probably have many of stored and ready to use, than to quickly start producing a new one.
In any case, the A7r3 hardly represents a 'mid-range' body, any more than the D850. The contest between it and the D850 pretty much comes down to AF - image quality, speed and other features are more-or-less comparable (3MP more on the D850, 1-2fps more on the A7r3, focus stacking on the D850, pixel shift on the A7r3, base ISO 64 on the D850, half a stop better high ISO on the A7r3). Any higher-resolution body is likely to be much slower - say, 70MP/5fps - with data bandwidth being the main bottleneck. The 'Sony being caught off-guard' bit is not with regards to the sensor being 42MP rather than 50-60MP, but with regards to it being the same 42MP sensor as the A7r2, without incorporating any of the sensor refinements of the past 28 months that would improve non-IQ aspects of performance - chief among them, stacked sensor technology. More than likely, they've already got a high-resolution sensor ready (at least design-wise - maybe not in terms of capacity for mass production). But that would make for a slow-shooting camera, not something to compete with the D850. They may not have had an updated 40-50MP sensor ready, that could shoot at the speed needed to compete. No-one expected the D850 to be a 9fps action camera.
