I managed to remove the Ch 1 attenuator and disengage the latching relay stack up. Applying 32V directly into the BNC connector and monitoring the hybrid output pin with a multimeter, there was definitely leakage and was measuring about 140 mV (140 nA across 1Meg). With the relays removed the only current path was through the input cap.
After I did some careful cleaning with 100% alcohol around the input cap, I remeasured and the result was worse. Measuring about 4V (4uA) leakage. I ended up removing the capacitor and tested the capacitor outside the hybrid and got the same results. So the cap was bad, which seems strange to me. The capacitance did indeed measure 22nF, so that was verified. The new caps should arrive sometime next week and we'll see how it goes.
Curious if others have experienced this behavior before.