BruCo scans (see here for the most recent one) showed a large coherence with IMC alignment signals, in the region between 100 and 1000 Hz. The most significant coherence is with IMC-WFS_B_DC_PIT.
I computed the transfer function between this IMC beam spot position signal and the OMC DCPD SUM signal, for a few different time periods in the past couple of days. The transfer functions are basically non changing over time in the different locks. This suggests that, if the increased coupling is due to a IFO misalignment, this is quite repeatable and constant over time (fig. 2)
Not much of a smoking gun, but some hints.