The basis for my reasoning is as follows:
First, we can group LOMC cartridges as having similar very low inductance in the 1 to 50 microHenry range. Typically more like about 10 uH for a very low output MC. So I would imagine that differences in inductance factor out, and no one is talking about using a less than 47K ohm load with an MM or MI type (and parenthetically, I agree with you and Ralph; I use 47K with LOMCs.)
Second, the effect to which I refer would have to do with parallel capacitance not inductance. Your quoted data re the DL103R (off by 0.5db at 20kHz) shows that the effect I cited is small, but do we know what were the capacitance values in that test? The idea is that impedance due to parallel capacitance will change from a high value to a lower value as frequency increases from 20Hz to 20kHz but even at 20kHz paralllel capacitance is not the dominant source of impedance (typically in the pF range). As impedance due to capacitance goes down with increasing f, then proportionate to impedance at 20Hz, net impedance goes down at 20kHz, shunting slightly more of the signal voltage to ground. Thus, compared to signal voltage at 20Hz, signal voltage at 20kHz would be reduced. Hence the potential for HF roll-off if cartridge internal R and phono stage input Z are mismatched to begin with.
