I was sending a similar PWM + LTZ1000 reference into production once. Though the application was different, I needed stable 5V from the 7.2 instead of 10, but the principle is very similar.
I'm missing something from all these schematic, which was very important in my experience. R30 and R31 (from the PWM10V1 schematic) has to be low tempco, big physically, and special care has to be done for the routing. C44 and C45 also. Foil capacitors to start with, with as small leakage as possible. Otherwise there will be problems, like leakage shifting the output, and the like. I even put a leakage guard trace (driven by a whatever opamp) below C44 and C45, protecting that entire node. All leakage gets multiplied by 100KOhm so thats why.
Otherwise the circuit is... a copycat. Not doing judgement, but there are better opamps available than back then. Also, the JFET switches