Wow that’s a pretty interesting (and scary) challenge.
I know that the coordinate systems get revised every few years. In Australia we used to use AGD83 then it was revised to GDA94 and more recently to GDA2020 — and this was to account for tectonic shifts. And there’s calculations for transforming from one to the other.
That’s the only clues I have on that side of things and I’ve forgotten most of what I’ve known.
Also there’s a concept called “rubber sheeting” for transforming points where you know the amount of error at a bunch of points and you want to transform all the other points by interpolating how much error there would be at those points. The might be useful too, if you’re implementing a solution yourself.
I didn’t find anything in QGIS to give a layer a CRS and an associated date, so I’m not sure how coordinate systems like this are expected to work properly.
I know that the coordinate systems get revised every few years. In Australia we used to use AGD83 then it was revised to GDA94 and more recently to GDA2020 — and this was to account for tectonic shifts. And there’s calculations for transforming from one to the other.
That’s the only clues I have on that side of things and I’ve forgotten most of what I’ve known.