Add a ledger readiness probe #37

User story / Problem statement

Currently the only health endpoint is liveness-only: it answers as soon as the process is up, regardless of whether the ledger is reachable. As an operator, I want a separate readiness signal, so that a scheduler does not route traffic to an instance that cannot serve a single request.

Expected outcome

A readiness endpoint reports ready only when the ledger answers, and unavailable otherwise. Liveness stays independent, so a ledger outage does not cause the orchestrator to kill an otherwise-healthy process.

Acceptance criteria

• A readiness endpoint returns 200 with a ready status when the ledger responds
• It returns 503 with an unavailable status when the ledger call fails
• The failure is logged, consistent with how other server-side failures are logged
• The 503 never reaches the terminal error handler, so a failing probe does not read as an unhandled server error
• The liveness endpoint keeps its existing behavior
• The suite is green

Technical notes

The probe body shape stays a status object to match the liveness endpoint, rather than funnelling through the generic error serializer, which would return a different shape and break the probe contract.

Additional context

Part of #22. Implemented by #18.