Hi Volt,
I want to start by saying the mics you have will make great recordings and what follows is more academic than practical, but it seems like you are open to more info on how matched your mics are...
A couple of things stick out when looking at your posts. You need to keep in mind there are two different aspects of mic performance that need to be matched for the mics to be a "matched pair".
1. Freq Response - this is the shape of the freq response plots of each mic. You are showing these in the captures you posted from Tombstone. One thing that is a little deciving in your captures is that you are using a 100dB scale. Almost anything will look good when you zoom out that much! To put these measurements in better perspective you need to use a much smaller scale so the differences are more apparent. Like 20-50dB. Each division in your graphs is 5dB. If you zoom in a little more you can see the differences more easily. The area of the curves you posted just before 6kHz with a near vertical slope shows a difference that is likey worse than the 1dB difference you are claiming. One easy way to see this is to click the L-R button at the top of the Tombstone screen. If you zoom in on the difference graph you should be able to see exactly how different they are at every frequency. The spec sheet for DPA SMK4061 stereo matched pairs allows each mic they make to vary from their ideal target response by +/- 2dB from 20-20kHz.
https://images.static-thomann.de/pics/atg/atgdata/document/specs/113026.pdf This means two random mics' freq response curves could differ by 4db at any frequency. Clearly the mics you have, being very close in response shape look better than this. Others have noted that DPA hints that the repose curves for mics with closely matched serial numbers tend to be much better than the allowable spec.
2. Mic sensitivity - this is the difference between the overall level of each microphone's signal when exposed to the same sound intensity. In the DPA data sheet I linked above you can see that the "Nominal Sensitivity" for any random mic can be +/-3dB from the target sensitivity for these mics (6mV/Pa). So any two random mics could be different in sensitivity by 6dB worst case. But please note there is a second specification for these matched mics in the kit - "Sensitivity Selection Tolerence". This is specified at +/-1.5dB. So they have improved the nominal random matching sensitivity tolerence by 1/2. This could be a big deal if you did not have an easy way to correct for this in post (bump up the lower gain mic channel by the difference between the mics). Fortunately most people around here are able to accomplish this easily with software so again this should not stop you from getting great results.
From what I gather, DPA is doing sorting similar to what happened with your mics. They are choosing mics with very similar freq responses (probably very close in the same batch) and then sorting further for close gain matching. Knowing DPA, they are likely doing MUCH better than the absolute worst case they are listing in the specs. I have not measured my SMK4060s but I would not be surprised if the responses and sensitivities were both within +/-1dB. This means that without any adjustments, their responses are within 4dB at all frequencies. Just a guess... Two random mics could be 10dB different at the extreme worst case.
You noted that you adjusted the mic gains between runs to get the curves to overlay better. This is due to the difference in sensitivities between your mics. Do you recall how much you needed to adjust the gain to get them to closely overlay? This would give you an indication of how far from a "factory matched" pair your sensitivities are.
From your curves, it looks like you did a nice job placing both mics in the same position when taking the measurements. Small differences in the position of the mics can lead to big differences in their responses especially at high frequencies. This is a major problem for people trying to take accurate mic measures at home. Making a jig like you described certainly helps a lot.
I don't buy into the thought that split onmi's don't need to be matched. Since your brain is still using both level and time differences to resolve the position of the sound sources, differences between mics will cause the image to be different than it should. If sensitivity and response matching didn't matter with split onmi's, shouldn't you be able to pull the level of one channel down or spin the treble knob for one channel in post and NOT have it affect the image? Try it out.
The great thing about split onmi's is that for distant sound sources, if both mics are matched, the image relys solely on the time difference between the mics created by the distance they are separated. Since the mics have an omni polar pattern the angle the sound is hitting the mics shouldn't matter. If the mics sensitivities are not matched, there will also be level differences fouling that up. As usual there is a whole spectrum of "how different" the mics need to be for this to really cause an issue. The further you separate the mics, the less this level difference can foul things up for sound sources that are at the far edges of the recording angle since the time differences are big. You would need a lot of level difference to overcome that (the closer mic much less sensitive than the far mic). But for sound sources that are in the center of the recording angle, there is no time difference between the mics, so any differences in level will shift the center image. A 1dB level difference leads to about a 7.5% shift in image position over most of the recording angle (Wittek & Theile).
Again, the mics you have look like a really nice match (freq response). If there are any sensitivity differences you can eliminate them in post and end up with matching that is as good as the overlays you created in Tombstone. Bet they sound damn good!
Miq