Which Talos API version streams enforce cancellation semantics for timed-out configuration requests?

Talos Linux exposes a declarative REST API for all cluster management operations, where request timeouts are primarily governed by the calling client (talosctl or custom integrations) rather than server-enforced limits. The talosctl command-line tool provides a --timeout flag for operations such as cluster apply and machine update, allowing operators to specify a maximum wait period; beyond this period, the command exits with a timeout status without automatically terminating underlying node processes.

Because Talos operates as an immutable OS, cancelled or timed-out configuration applies may leave the node in a pending-update state until the next successful API poll or reboot, a behavior documented in release notes but not uniformly addressed in all version streams.

Which Talos API version streams enforce consistent cancellation semantics for timed-out requests? Do exit codes and node state transitions reliably correspond across talosctl releases? How does the declarative API surface timeout failures to downstream integrations without requiring custom polling logic?