I am glad, that we agree. A lot of things interact here. The motor got a five pole rotor as well as the tacho generator (other small coreless DC motor with five poles of rotor). The 5 pole motor will normally have 10 preferred positions each revolution due to this cogging moment. But motor also got some 10 times variation of driving moment, and the tacho generator also got a ripple frequency 10 times revolution frequency. Some unilinear mix of all this seems to make this 20 times each revolution.
However this cogging frequency do also represent a rotational resonance, that can be hard to counter by a controller, because you can easily get instability in the closed loop at this frequency. I did also see a rotational resonance in the mechanical connection to the tachogenerator. All these mechanical resonances makes it harder to have a closed loop control of speed with high gain.