It may not be a big deal depending on how HEPI is used but the Inductive Position Sensors are close to their limits.

With Keita applying -12000 somethings offsets in the Blend for Ry, V1 & V2 are close to 30k counts on the IPS.  I increased this to -13000 to check that we still had some headroom.  First attempt tripped as the IPS watchdog is set at 30000.  Why isn't this 32000?  I increased the trip level and went to -13000 again.  All the IPS moved in response and the spread in readings was around 10% of the change so it didn't seem any one actuator was stuck.  Looking at the Dial Indicators though, the SW corner (2) moved down 6 mils while the NE corner (4) went up only 1mil.  The other corners shift 3mils each.  The magnitude of the moves make sense, the DIs average 3.25mils & the IPS average 2.76mils.  But the IPS are much more uniform than the DIs...  Could the V4 be hitting a stop, hard to tell but it didn't look closer than the others.

While at -13000 Ry I checked the Actuators and they all looked pretty close to hitting their mechanical stops.

Removed the -13000 offset and checks the DIs & Actuators.  The actuators looked decently centered but the South side rebounded up more than the North side rebounded down by ~2x. There was also several mils of Rz displacement seen at the DIs.  Returned the -12000 offset an the IPS return to almost exactly where we started.

Couple things:

Much further actuation and possibly damping/isolation coud trigger the IPS watchdog with the current 30k limit.

Correcting the input matrix, installing the symmetrization filters and commissioning the loops would be good to get done.