How could i mesure the peak frequency of a moving nonresonant analog filter without a feedbackloop between its audio output back into its audio input?
I do have a calibration patch for resonating filters and oscillators.

What would be a good way to do
this? White noise into the filter > FFT > Findung the peak? And slowing the whole calibration procedure by the overall latency. Or better with a sine sweep instead of noise for being independend of randomness.

How can i find such a peak?

have a look at H13.butterworth.pd and the filtergraph1 and filtergraph2 abstractions in the audio examples.
and you'll need to delay the reference signal (2nd & 3rd outlet of filtergraph1) according to your systems latency. I haven't tried this with analog filters, but i guess it should work..