Next, I would use a precision granite parallel shimmed to dead flat on the chuck and run a micron indicator over at at long and cross travel to test a warmed-up machine, this to see if the machine runs dead flat.
One might fresh dress then take a full wheel pass one a sacrifice part also set on the chuck before grinding the finish pass on the needed part, and come to the part with only touching at less the full clean-u / making perhaps three grids to full clean-up.
On a stout part like 1"x4" I have placed two shims under at near the ends (about at the thirds) so to eliminate chuck flatness error, yes the half or so mag helps when doing this.
You might give wheel specs and part hardness so the guys can add more suggestions.
I have had parts that needed zero error and have used a lapping block when that took less time than careful grinding.
I have steel flat parallels that I lay a part on when I don't have time to grind a whole chuck . They are grooved on the top so heat cant build up at truing.I just tickle grind these and and de burr with a few swipes of crocus cloth, then set the part on them for grinding the part dead true.
OT:
Good you asked, my answer adds 421 words to the book I am trying to write.
Yes go back and read this over because I am still adding things as I think of them./ look at the count.
Very important that all the grinder guys keep pitching in other ideas on this problem.