It's only marginally on topic, but this stuff should probably be addressed:
The second part (running the way you like) definitely should be taken into account a lot of the time, but the bit about pure physics and even pace really only applies to some theoretical machine that could draw from its various energy sources at an even rate throughout and would never have to allot any additional energy to accelerating at the start. Of course, a body isn't that type of machine. Its energy pathways don't always operate at a constant rate; nor do they interact at a constant rate. When running at a reasonable 800m speed, in the act of accelerating to speed off the starting line - and when running at race pace (even a second or two per 400 slower than 800 pace) - the lion's share of contribution from anaerobic pathways (phosphocreatine and glycolysis) occurs pretty early in the race. The overall energy distribution is 55% to 65% via aerobic metabolism (closer to 55% for a race of shorter duration - i.e., 1:41 as opposed to 2:30) and 35% to 45% anaerobic, with the contribution from aerobic metabolism rising continually throughout the event. The best strategy with the kind of speeds involved in the 800 is to recognize the need for some sense of pacing, but to get a faster-than-average-pace start and make the most use of the anaerobic pathways early, while they're delivering their peak contribution to the effort. Get the ball rolling, so to speak. In this way, you can almost cop a "free" second or two versus trying to run at a perfectly even pace. Are there some runners that might be the exceptions to this "rule"? Yeah, sure, maybe. Those are probably the guys who aren't too fast at 400 and can't run the opening lap very much ahead of pace without being maxed out early and tying up like a big dog.
So it even makes sense in theory to try for a slight positive split in a PR attempt at 800. Of course, new measurement techniques will arise and there will probably be future research that calls into question or even refutes outright the findings of all that research that was done between 1994 and 2006, which primarily used the accumulated oxygen deficit method and measurements of changes in muscle substrates to assess the contribution of the different energy systems during various events. But at the end of the day, you know what? Theory is buggar-all. Actual performances paint a better picture than small-sample lab research ever could. Sure, a number of championship 800m races have been won with even or negative splits, but - despite all the attempts to tinker around with record-setting 2-lap strategies over the decades - only the tiniest number of career-best times are achieved this way. If records continue to fall, some people might conjecture that there will be a tendency toward even splits. But I'm guessing that ideal splits will still be positive (though perhaps slightly less positive). Why? Because if 800m runners as a whole get faster, it stands to reason that more of the pool of 800 specialists will also be faster at 400, allowing the vast majority of them just as much wiggle room between top-end 400 speed and average 2-lap pace as people have now. They'll still be getting out ahead of average pace without having it be too fast. Will there be exceptions to that - people who run their best evenly or possibly with negative splits? Yes, but there always have a been a scant few.
Your body starts almost exclusively anaerobically and - while energy-supplying processes overlap and interact - it takes time for the aerobic energy production to reach its maximal contribution, but when it does, it dominates. In an event as brief as the 800, peak contribution from anaerobic energy pathways occurs rather early in the effort and continues to decline throughout, while contribution from aerobic metabolism rises, although VO2max is not reached in such a short time.
On a geologic time scale - i.e., by comparison to when the Burgess Shale was deposited - yes, that research is brand-spanking-new.