Here's an analysis of those injections. In brief, the coil and mechanical responses are the same for ETMX and ETMY as expected. The differences are all due to filters in the L2 LOCK and DRIVEALIGN filter banks.

First plot shows a comparison of the L2 LOCK L EXC to CAL-DELTAL transfer functions for ETMX and ETMY: they look different.

In the second plot I plot the filter chain from the L2 LOCK EXC to the COILOUT. They are different, probably because ETMX is used for DARM control, and ETMY is in some legacy configuration (note that Camilla turned off the LPL2L3 filter). The third plot shows the ratio of the ETMY  / ETMX transfer functions as measured, and what we expect from the different filters that are engaged. They match very well.

If we compensate the EXC to CAL-DELTAL transfer functions with the LOCK and DRIVEALIGN filter, and take the ratio, we get a transfer function pretty much at 1, as shown in the fourth plot.

Finally, I calibrated CAL-DELTAL_EXTERNAL in meters using Louis' file from /ligo/groups/cal/H1/reports/ and the fifth plot shows that the DARM / COIL transfer functions are pretty much the expected 1/f^2, with only a sign difference between X and Y.

In conclusion, there isn't anything wrong with ETMY PUM drive. More thinking is needed to understand why the LSC FF have a phase rotation much that is much larger at LHO than LLO, and how to better tackle that problem.