There is no reason a load developed at 100 yd will not perform well at 600 yd. I've done load development at 100 yd for years and the resultant loads shoot just fine at 300, 600, and 1000 yd. Further, shooting 5 x 5-shot groups during load development will likely ensure you burn out your barrel long before it should have been necessary. If you're confident in your ability to shoot solid groups, a single 5-shot group will tell you all you need to know. Certainly you have to pay attention while shooing your groups; if you noticeably pull a shot or let one go when the wind condition changes markedly, write it down for later use in evaluating the group.
There are several potential issues with the data you presented above. First and foremost, it is more challenging to shoot with good precision using an AR-type rifle due to the movement of the BCG (as compared to a typical bolt gun). If you can turn off your gas system and single feed during load workup, it might help. Second, was 23.0 gr N140 the only charge weight you tested? I'm curious as to why the velocity decreased as COAL decreased, yet the ES increased during the same seating depth intervals. My gut feeling is that you have not tested a wide enough window during charge weight testing to find the optimal charge weight window. Changing the seating depth of a jumped bullet by .010" shouldn't alter velocity by that amount, although it might for a jammed bullet, or one seated very close to "touching" the lands. If you find the "sweet spot" during charge weight testing, i.e. a window of at least two or three consecutive 0.1 gr charge weight increments that give you minimal velocity increase as the charge weight increases AND give you low ES/SD, that is where you really want to be. At that point, testing seating depth from around .005" off the lands to at least .025" off the lands in .003 increments is what I would suggest. Look for at least two or three successive seating depth increments [window] that produce the best 5-shot groupings. Then you load to the longest COAL of that seating depth window, which will give you the most headroom to account for land erosion before you need to do another seating depth test.
Some of the keys to precision reloading are to test a single variable at a time, covering a sufficiently wide range to find and define the "edges" of the optimal window (so you can lad to the middle of the window), and to use sufficiently small increments so as not to potentially miss something. For a cartridge as small as the .223, I would carry out the final charge weight testing in 0.1 gr increments, and I use 0.003" increments for seating depth testing. I think if you carry out your charge weight and seating depth testing in a rigorous and systematic manner, you will find a stable load with good precision. Good luck.
