Harden cluster controller to not panic on error

[tillrohrmann]

If any of the cluster controller components encounters an error, it is usually propagated to the service and then to the task where it causes the program to panic (example:

restate/crates/admin/src/cluster_controller/service.rs

Line 330 in 57625b3

.await?;

). Instead of panicking, we should handle the error graceful.