Give outer cron deadlines headroom beyond inner work budgets

A strict one-shot scheduler timeout must exceed the application's internal processing deadline plus container startup, name resolution and TLS, proxy routing, response serialization, and cancellation margin. Otherwise ordinary routing variation can make the caller exit nonzero even when the server later returns success with durable partial progress.

Order budgets explicitly: inner work deadline below route or platform deadline, which remains below the scheduler deadline. Emit phase telemetry for each boundary and make retries idempotent. Increasing the outer timeout can prevent false crash classification, but it does not fix slow work, unavailable dependencies, or missing schedules; diagnose those separately.