Technology, as you point out, more broadly deployable, as well as cheaper and less likely to discriminate. For that reason, I actually kind of like speed / red light cameras. However, they are wildly unpopular to the point of being made illegal in many places. Plus, I think there are legitimate concerns about surveillance aspects of our newly smart cities.
With respect to "safer at any given speed" designs, I think you have to consider that, in order for safety to hold, you need to make some unrealistic assumptions about other factors remaining constant. Also, many of those so-called safer designs really do everything possible to eliminate other kinds of roadway users from the mix.
An automobile monoculture isn't very robust, and it also assumes the wrong units of analysis. Traffic engineers used to measure mobility in vehicle throughput. Amazingly, if you shift your units of analysis to people, monoculture looks less compelling. And that's even before you consider physical activity, environmental externalities, and a raft of other issues.
But this is going to be a hard sell for both drivers (who feel entitled to their high speed) and law enforcement (which treats ticketing as a revenue source; widespread deployment of such system would dry up the money stream almost immediately).
It wouldn't be a tough sell to lower speed limits on some portions, and raise it on others, even on a time-neutral but safety-positive basis, but yet, the powers that be stick to their arbitrary number.
I haven't seen any actual calculations used for setting speed limits, but I do remember being taught in school (on physics lessons) that if you want maximum road throughput then after factoring in safe distance (which is proportional to square of velocity) you end up with ~50km/h as a sweet spot. Even if some other factors (like fuel efficiency) alter this calculation, 100km/h feels quite plausible to me.
Also I believe that many consider speed limits as arbitrary because most people seem not to care about safe distances anyway.
Or change the context in some third way (e.g., enforcement)? The thing with roadway design is that the roadway designs which enourage higher speed are generally the same ones that are safer at any given speed, so changing the design to reduce the incentive for high speed may not improve safety as much as the reduced "natural speed" would seem to suggest -- and may retain the high delta beteween fast drivers and slow drivers that you note is a major factor in accidents, just reducing the median.
But what if through technology we could get enforcement of existing limits to a place where the probability of penalty for violations was higher (even if the penalty range had a lower bottom end)? One problem with law as an factor in behavior right now is that the probability of any given violation (or even a pattern of consistent violations) being punished is so low that the law serves more as a very weak social signal and the source of an inverted lottery of punishment than as a rational utility factor in regular behavior calculations.
With more consistent enforcement, we may be able to reduce the delta between high and low speed drivers (as the high speed end is usually significantly above the legal limit) without changing the law (at least, the speed limits in the law) or the road design significantly.