Ah, the truth comes out! Someone just tapped the RCA outs and brought it to the front and called it done. This is wrong as the RCA out typically has high impedance in the order of 100 ohm we see here. That high impedance causes a ton of power loss as we see in the 33 ohm test.
Have you measured the output impedance? It's not in the review. According to the manufacturer, it seems to be 10 Ohm. However, it is not specified for which output.
Source: https://1mii.com/wp-content/uploads/2020/12/DS600-Manual-12.24.pdf
Further, the measured power at 300 ohm of 13.26 mW corresponds to the line-out level - 1.994 V for channel 1 giving 13.25 mW for a 300 ohm load - suggesting that the output impedance is near zero. A 10 ohm output impedance would instead give 12.4 mW.
The power loss at 33 ohm being entirely due to the output impedance would require the output impedance to be about 133 ohm. Thus, the output impedance must be highly load-dependent to allow for the measured power at 300 ohm at the line-out level.
For the 33 ohm load, the current corresponding to the maximum power is about 15 mA. At 300 ohm only about 4.7 mA is required to reach maximum power. Since there is no rapid rise in distortion after maximum power, which is reached at only about 25% of the line-out level, there seems to some load-based limiting before the output stage - or the output impedance is indeed highly load-dependent.
Thus, if it were the case that the output from the RCA and the headphone output are the same, the output impedance must be the same on both and their power and distortion must be the same for the same load.
Inasmuch as you have not so far, could you thus not only determine the output impedance of the headphone output but also that of the RCA output and further, determine the RCA-outputs power and distortion at 33 ohm and possibly also at 300 ohm?