I have not seen behaviour like this from my own Maynuo 9712, and any power supply in my posession (switched or linear). When switched off, (or when on and the output dissabled) the outputs appear as a reverse biassed diode. This is a function of the main current sinking mosfets. When forward biassed (again, whether switched on or off, but with the output off), the foward current is so totally small as to be negligable.
The positive and negative poles of the Maynuo are not connected to ground (floating), though there is around 20nF capacitance to ground.
Depending on the testing frequency, I measure between 4 and 2 uf input capacitance and 400 down to 10mH of inductance. So there might be a slight resonant load, there. This may be my LCR meter confused by the diode effect of the load, when the output is off.
This same (mildly) confusing load may be having an effect on your SMPS, too. But to be honest, I think it is a design weakness in your supply, which ought to be stable enough to deal with mildly reactive loads such as this. A ferrite bead on the output should quieten things down.
Try a battery of a similar voltage across your Maynuo and look at the waveform - anything?