Well stated. This corresponds exactly with my experience with spectros and older colorimeters. They all age but colorimeters age faster. Worse, colorimeters XY white points drift far more and that's a killer. Spectros tend to just drift lower in luminance and xy (color tint) drifted much less.
This is really the most important point here-- A colormunki being a spectro with a grating isn't going to shift color as badly since it's not relying on interagreement of multiple different color filters. I would say you're well into the world in which scientific precision begins to break down and some element of subjectivity needs to be taken into account. You'll never have a perfect match of any screen to any print, or any screen to any other screen using a different backlight/phosphor technology. IMO, if the grey tones look tinted, change them until they look good to your eye and calibrate for that. Or pick a paper you'll be printing on most of the time and do a visual match of the white point. But if you're using your monitor for doing print proofing, video color grading, DICOM viewing, etc. all at the same time, then you've just got to pick a happy medium (or be insanely diligent about switching built in LUTs).
I'd keep your current spectro, but add an i1displaypro3 or colormunki display, and use the spectro to create a correction matrix for the colorimeter using displaycal. Colorimeters are MUCH better at seeing detail in dark tones (SNR gets too high for spectros to accurately read near-blacks in emissive/transmissive modes) but your spectro will be more accurate at obtaining accurate values for color primaries, as explained by Pat earlier.
besides getting better at reading dark values, the only way I can see spectrophotometers really "upgrading" in the near future would be to decrease the space between the grating steps, therefore measuring in higher detail to deal with very spikey spectra, but a better solution for that would be for our display technology to just get better at providing a smooth response.
There's a post on AVSForum talking about the recently released i1pro3 and how it basically doesn't improve upon the previous versions for profiling displays. Also see:
https://www.argyllcms.com/doc/i1proHiRes.html