Tagging the SEI, SUS, AOS, ISC, and SYS group, since the majority of these channels are projected versions of the ITMY ST2 GS13s and I suspect solving the problem will require action from a broad number of groups.  

~74 [Hz] is roughly the right frequency for the fundamental of QUAD cage resonances. For H1 and L1 ITMY specifically, See LHO aLOG 5322; The X (beamline) direction attachment shows a broad resonance at ~75 [Hz]. Similarly for LLO LLO aLOG 7214, though the Q of the resonance is significantly lower.

Suggested mechanism for coupling to DARM: the QUAD's cage is attached (mechanically) to the the ISI-ITMY's ST2. Thus the GS13s (the most sensitive local sensor in the chamber at those frequencies) are likely to see these resonance, where the resonance is excited by residual ground motion. However, the QUADs are great isolators at least frequencies, so we can rule out "traditional" coupling via direct transmission of this residual motion through the suspension chain to the test mass. What's more likely is that the cage itself is picking up scattered light from some source, and coupling it directly into the cavity's beam line. 

Note that a similar frequency, low Q feature was found on H1 ETMX while our end-station beam diverter was inadvertently left open for the beginning of the run, see LHO aLOG 22350. I suspect the same mechanism is at play here -- scattered light glinting off the QUAD cage into the IFO beamline.