@leftwo was working on a Sidecar and had commanded the system go to A2. While it did successfully do so, it left all the front I/O LEDs in the last state that they were in when dendrite had operated it. Given that the switch zone and dendrite only generally expect to operate while Sidecar is in A0 (though this may be something we need to consider and change), we likely should be defaulting to turning off LEDs when in A2.

This definitely requires some deeper thought about how we want the service and control points to operate while we're doing troubleshooting and in-field service, but given that this can happen independent of dendrite existing, they likely shouldn't remain on. I expect the flow of control and the ordering of different requests via the broader control plane and dendrite is something that will still want to continue to evolve, but at the moment what we have for behavior is rather confusing.