Bound ciphernode startup with a configurable readiness deadline #1652

Audit reference: I08

What can go wrong

Ciphernode startup could wait forever if a subsystem never reported ready, leaving a hung process that looked alive but could never serve work.

When it happens

This occurs between process launch and readiness, for example when RPC, networking, storage recovery, or an actor stalls.

What changed

Startup now has a validated, non-zero configurable deadline and returns a typed timeout error.

Why the change works

The service manager gets a definite failure it can restart or alert on instead of an indefinitely half-started node.

Checks

Default, zero-value rejection, successful readiness, and timeout tests pass.