repro

1. start the cluster in secure mode (with login turned on by default, see PR #TODO)
2. go to the admin UI

expected

Be directed to a login page where I can log in and proceed. (if we had one)

actual

Healthchecks fail with 401 unauthorized, showing the red "connection lost" banner even though the connection is fine; we just haven't logged in.

solution

Exempt the healthcheck endpoint, /_admin/v1/health, from authentication.