I agree with everything you have said. MSFS data is wrong, as all VORs seem calibrated/aligned to the current mag var, not the historic value at the time of installation.
The one point I disagree with is that published courses are never amended. You are 100% correct in that the published course is the number of degrees from the stations 0° reference. If, over many years, the mag var for the site’s location changes by 5°, than the magnetic radial required to draw the same track over the ground now changes by a corresponding value. Whilst the 090° radial will always be 90° offset from the reference vale of 0°, the magnetic course required to fly from A to B will change over time. Given these are the values published on IFR charts, they are amended over time to ensure the published value will still provide the same ground tracking.
I guarantee that if you viewed the 1965 charts for the above discussed navaid, the published magnetic course for airways emanating from this VOR, would be different to what they are on today’s chart.
This leads to the problems in MSFS. All navaids are aligned to current magnetic north. Therefore any published course that has been corrected over time to account for the different between the navaids installation alignment, and the sites current mag var, actually introduces an error, rather than removes it. IFR charts effectively become the source of the error as their corrections are no longer require.
If the Navigraph data aligns VORs to the sites original value, rather than the current mag var, as you suggest (I hope you are correct),than current IFR charts become correct again.