It might be helpful to refer to Section 2-34 "Insignificance of Inherent Meter Offset" on page 2-5 of the 8050A manual.
RMS converters produce a DC voltage proportional to the RMS value of the input. The output will always have a small DC offset due to circuit imbalances. This is an error term which becomes less and less significant as the applied input voltage goes up. For that reason it is not appropriate to use the relative offset feature (REL) to get rid of the error. It is also why accuracy is specified from a certain percentage of full range, up to full range. Fluke is saying 5% for this lower limit, which basically means at least 10mV on the 200mV full-scale A/D converter should swamp any errors.
It was a mistake on my part earlier to use 750V as the full-range value, because that's basically an imposed safety limit. For the highest range, full range is mathematically 2000.0V, even if it might arc over or exceed component ratings. 5% of 2000V is 100V, so to stay in band, it would be 100V and up. Likewise, 10V and up for the 200V range.
From my DIY/hobbyist experience, the meter will be sufficiently calibrated for mains-frequency AC on all ranges with the two interacting potentiomenters R7 and R29 and the 100mV and 1.9V test inputs. The tolerance and stability of the voltage divider resistors should take care of higher ranges.
At low frequencies, the resistive components of the voltage divider dominates its impedance. But as frequency increases, reactance components from stray capacitances start to become non-negligible. The C1 and C2 trimmers are there to compensate, and that's why the procedure calls for 10kHz inputs for their adjustment. I do recommend leaving those alone.
Your 8050A is actually within spec at 5VAC and 110VAC on the 20V and 200V ranges, assuming the 8800A is golden, which of course is not guaranteed. It still might be interesting to try checking at the 100mV and 1.9V AC levels. A little tweak might improve the "downrange" readings. And for the RMS converter, check the reading with shorted inputs in the 2VAC range, the book says it should be less than 40 counts from zero. Mine is 3 counts from zero (0.0003V).