After the OAF model wiring fix, the Suspoint -> IFO-basis appears to work fine, except for the (already mentioned) limited subtraction.

I compared the GS13 (Suspoint) DARM with H1:CAL-DELTAL_EXTERNAL_DQ. To do this, I had to de-whiten the DELTA-L signal with 6 zeroes at 30 Hz 
and 6 poles at 0.3Hz (10^12 dewhitening). In principle, this should be calibrated down to 10mHz (the calibration model has a pole at 10mHz 
instead of DC). 

The two spectra (attachment 1, 'DARM_suspoint_vs_IFO.png') somewhat agree. There seems to be some residual error in the calibration of 
Delta L at low frequencies (it should cross 10^-6m/rt(Hz) at some point). Kiwamu found that the gain field of H1CAL-CS_DARM_FE_ETMY_L1_LOCK_L 
was a factor of 3 different from its equivalent suspension model, and corrected it, but this cannot explain the difference we see. 
The coherence (attachment 2) is quite good from ~0.03-0.3Hz.

Note that the GS13s are calibrated only to 30mHz, but this is also where they begin to rapidly lose coherence anyway. GS13 noise floor 
(for a single device) is added to show that there is some significant SNR at nearly all frequencies shown.