Activated bounded HOS failure triage and self-recovery.

A control-plane workflow now classifies eligible failures and can perform one bounded recovery attempt for retry-safe workflows before escalation.

What changed
01

Adds automated failure triage for supported control-plane workflows.

02

Limits recovery to a bounded one-time retry where the workflow is explicitly retry-safe.

03

Keeps escalation available when recovery is not appropriate or does not resolve the issue.

Why it matters

The product can recover from a narrow class of known operational failures without turning failure handling into uncontrolled automation.

Release-note boundary

Shipped means verified, not planned.

This update records work that had a verifiable release or active control-plane state at the time it was published.

Roadmap ideas, experiments, internal discussion, and unverified capability are not presented here as shipped product.