nexus: allow instance updates to complete when one switch is down - #10389

control: job assigned to worker 01KR1Q658KSEKX9XH035EMK3RZ [factory aws, i-003eee79de33b10bd] (queued for 53 s)
...
task: starting task 0: "setup"
task: process exited: duration 4102 ms, exit code 0
...
task: starting task 1: "rust-toolchain"
task: process exited: duration 15949 ms, exit code 0
...
task: starting task 2: "authentication"
task: process exited: duration 16 ms, exit code 0
...
task: starting task 3: "clone repository"
task: process exited: duration 10444 ms, exit code 0
...
task: starting task 4: "build"
...
|  Documenting repo-depot-api v0.1.0 (/work/oxidecomputer/omicron/sled-agent/repo-depot-api)
|  Documenting omicron-workspace-hack v0.1.0 (/work/oxidecomputer/omicron/workspace-hack)
|  Documenting omicron-uuid-kinds v0.1.0 (/work/oxidecomputer/omicron/uuid-kinds)
|  Documenting dev-tools-common v0.1.0 (/work/oxidecomputer/omicron/dev-tools/common)
|     Finished `dev` profile [unoptimized + debuginfo] target(s) in 6m 30s
|    Generated /work/oxidecomputer/omicron/target/doc/api_identity/index.html and 194 other  [...]
| 
| real     6:30.635578653
| user    17:23.679538467
| sys      3:41.942218562
| trap        0.254939563
| tflt        0.108268080
| dflt        0.046009465
| kflt        0.000027640
| lock  6:05:43.987555880
| slp   2:11:18.751891148
| lat        38.543115689
| stop     1:44.038456648
task: process exited: duration 779302 ms, exit code 0
|W| found 0 output files