I found one problem, but am unable to diagnose the transient I am seeing around what looks like the oven going on and off.
If you find your are out of spec at the 0000 calibration step, investigate R24. I added 1k in resistance to that location and it brought my readings from +30mv down to +1.2mv. I have dialed that 1.2 down to 0s and am waiting on the unit to stabilize before dialing it in further.
As far as transients, has anyone ever investigated their cause?
This unit had 4 problems, these are the mitigation steps taken.
1. The failed calibration with R24
2. Too much ripple solved by recapping
3. Voltage drift upwards, this was resolved incrementally from all steps but benefited the most from fix 4.
4. I spent several hours digging into the issue with transient voltages appearing on the output. The supply is supposed to stay stable at .0000, this one was doing that but had a lot of variance past 4 zeros, so 0.0000.001 to 0.0000.999. When the transient appeared it was sending the signal up to 0.0002.xxx every 20 seconds or so, some instances more intense than others. It also continued to voltage drift upwards with 0000 on the dials (no output).
I tested the transistor paths back to the power supply, saw the transients everywhere I put the scope leads:

The transients were even across pins on the power supply itself, which got strange as I thought it was the power supply that was the problem. I tool a look at the schematic again and noticed the heater circuit goes right from the 115v AC source, which maybe explains how the transient was seen everywhere:

Up until here I didn't realize the tap for the oven was at the transformer. I looked at the 2 caps across pins 9-11 on the oven, C9 and C14. I changed both caps and added 2 resistors to form more of a snubber on the feedback. This worked, I ended up with 270 ohms, 75 ohms, and a 10nf ceramic cap, and 2 x 10nf to make the 20nf cap.
I still get the transient, but after several hours in stable conditions, the transient (from cold start) maxed out at 0.0000.400 versus 0.0002.000, a 5 fold decrease.

In summary, at rest, powered on, the unit is much more stable at 0000s, spends most of its time at 0000.010 or below, and after being stable 12 hours shows only two transients in a 15 minute test exceeded 0000.010, one hit .039 and the other .024.
Moving on to the 20v calibration, hopefully the stability translates there...
Edit: its a 220 ohm resistor, not 270.