So much of this verrrrrry long article at the link below parallels the projects I’m in the midst of in demixing a bunch of 1965-1968 projects.
In some ways, I’m gratified to read the article because I had to “make up” similar steps when I began the projects with initial tape transfers in 1982. I too, hired a Cedar retouch system in the 90s for parts of my projects…small world…it’s nice to read the journey of the Beatle camp team…none of us anywhere on the planet…Beatles or otherwise…had a comprehensive guide for demixing effectively.
And the techniques are still in infancy…even with current software.
I’m currently dealing with a few dozen hours of just pulling hihats out of mono drum tracks…matching ambience around snares that are pulled out…manually finding/replacing thousands of individual hh hits…which only sounds correct IF I do it manually…same stuff that Jackson’s team claim they’ve effectively automated with their machine models…which I’d looovve to be able to test myself 
One of the 3-minute pop recordings I’m working on from 1966 (which was originally tracked in four slave bounces at the time on several quarter-inch 15ips reels) is now represented as 6gb of components on the ssd! The track was transferred a few years back…speed corrected with Capstan to eliminate wow/flutter (and force it back to correct pitch), hum reduced (hiss is intact…I don’t trust removing it during the creation of new multitracks…tempo corrected (just a little) and chopped/demixed into a billion pieces …6gb! I know I’m not alone now
Jackson’s team still admits being unable to demix a mono mixed Rick 6string from a Gretsch 6120 without artifacts…good…I can’t either…yet At least they’re in the same boat as me on some issues.
At any rate, the article is a very long read…but infinitely interesting if you’re on the same trip as I am all these years…