During an access break to the ESR, I made a few tests with the bias scheme to see if we could better understand some of the weird quirks.

Two detectors showed likely shorts during initial setup - these detectors were still cabled to the bias network though. 
If there is a short to ground, we wondered if that could then impact the other detectors on the same MHV4 module.
These detectors were uncabled at the pre-amp end.
No change was noticed we bias was applied again.

On the outside pocket, I also noticed that when one detector is turned off, the leakage currents of the others remain the same amplitude but flip polarity as the other biases down.
I disconnected all detectors from that MHV module and connected them one-by-one and this effect was immediately seen with two detectors connected.
I double-checked the cabling and cannot see anything wrong there.