Great overview. My (personal ) experience from working in a research and innovation centered environment is that the evolution will go towards smallscale production as opposed to prototyping and new materials will be an important part of this, we already have resin materials that can be used to make injection molds for other plastics and I recently saw a desktop metalprinter in development, also a wider adoption in research areas other than tech related, will make it more mainstream. We often forget that it is slow going from the point where the early adopters grasp the potential, until it becomes mainstream, and it needs to become mainstream for the mature and cheap products to come out. I think thats what we are beginning to see with the lowcost resin printers.
If i should point at one revolutionary development (but i dont think that will be this years , or the next) it would be true replicators that is machines that can make other complete machines (including wiring, boards and motors). it is already technically possible but it has not been combined in one machine yet. I would love a printer that could print printers, and some researcher believe this is achievable within a few years
Yes, smallscale production seems to starts in few different areas.
If I compare it to the computer industry we are still at the end of 70's, we do not yet really grasp it, we do not have yet killer applications/products implemented with the 3D Printing potential.
I think too that we will see more MIM style desktop printing and the cost of a small electric kiln is not so expensive especially for the size of parts many people are printing.
What I am very interested is materials and multi-materials for metal/resin/plastic, it can enable crazy stuff for many applications.
One of the next step as you said will be the assembly, there are plenty of small bots (pick and place machines, "generic" bots) that are doing one or few tasks, the cost starts to reduce but most of the machines are not yet in the price tag of individuals or not having enough precision/force.
It will take some time also to have dirt cheap servos + fast controller with enough precision for this, the technology is already there but there is not yet the demand to really push down the price: may be the next generation FDM 3D Printers will push forward in this way. There is also the software that is not really accessible and a major bottleneck at this stage.