Hi Mike,
sorry, I did not want to offend you - just wasn't sure about your status of experience :-)
By the way, my own status is "interested - but far away from expert". Just jumped in answering your question as none of the more experienced users did...
To mention it first:
a) I don't use i1Studio
b) I have no first hand experience with DisplayCAL
c) I own a NEC PA271q, so I am a little experienced with hardware calibration.
That said, here some ideas:
1) The PA272W is capable of hardware calibration, AFAIK. Please forgive me if I tell you something you already know, but i1Studio/i1Profiler can not address the monitor's LUT directly, so they write their "corrections" into the video card's LUT, thus giving away some of the monitor's potentials (regarding gamut, blackpoint etc.). Thus I would recommend using NEC's Spectraview II or BasICColor Display to address the monitor LUT directly - or at least use NECs MultiProfiler software to set the monitor to the desired values.
But assuming you have good reasons to use the profiling packages you use, I would at least recommend using MultiProfiler to check the status of your NEC after resetting (of course you could use the on-screen-display as well, but MultiProfiler provides better overview). Here I would check for gamut, blackpoint, whitepoint and luminance to be set to "native" before profiling with i1Studio/i1Display.
2) Given that your NEC is in the same status as it was when using your old machine, I would check whether you had access to a profile made before switching to the new MBP2016. It should work, at least better than the blueish one you got now. If not, consider a hardware / connectivity problem (you could test it by using a different connection type and/or additional USB connection).
3) AFAIK DisplayCAL is not able to address hardware LUTs directly. But apparently it does a better job handling the situation than i1Profiler/i1Studio.
I can imagine this is due to using a LUT: As I can see from the screenshots, DisplayCAL uses a 100/010/001, aka "empty", matrix. In my understanding that means that DisplayCAL's matrix does "nothing", the actual correction is done via the additional LUT (quite similar as it would be with a profile written by SPII/Display - it's LUT would yet be stored in the monitor, not the video card). If, for example due to connectivity problems, a non-linear, strange behavior would occur on the NEC, the LUT of DisplayCAL's profile could perhaps correct that (better) than i1Studio's profile with it's linear correction matrix. Just a guess.
4) Looking at the profiles you uploaded, I see that "Media white point, Illuminant-relative XYZ" in DispCal's profile is 5174K, in i1Studio 6500K. i1Studio's PCS-relative XYZ is 5001K. Quite a difference... but I lack of proper knowledge about it's implications :-/
PS: Concerning my PA2711 I finally came to the conclusion that MultiProfiler alone was sufficient for me - it relies on the monitor's internal calibration checks, but they produce very similar, if not better results than achieved with external software and - comparably cheap - pucks.
Hope something helps!
Best regards, Michael