/readyz waits for the database, and /healthz is gone

/readyz now answers 503 while the server's database does not answer within a second, so a load balancer stops sending it requests, and the console's System screen reads it. /livez still answers while the process serves HTTP: point liveness probes there, so a database outage does not restart every instance.

From 0.2.0, zitadel start waits until the local server's database answers, zitadel doctor, status and setup count a local server as healthy only then, and withZitadel from @zitadel/testing waits for /readyz as well.

Breaking: /healthz is removed. Move liveness probes that use it to /livez, and readiness probes to /readyz. A zitadel CLI or @zitadel/testing from an earlier release waits for /healthz, so it does not see a server from this release as up: update them together with the server.