Worker/workflow-level hook to react to every workflow's terminal outcome (success/failure)

Is your feature request related to a problem? Please describe

We run a set of YAML workflows and need to react to each workflow's terminal outcome (success and failure) to update an external system — record status, the failure reason, etc. The constraint: we want this generically, configured once, not by adding a try/catch + status-report step to every activity in every YAML.

There's no built-in way to run logic on a workflow's terminal outcome across all workflows without per-activity/per-workflow boilerplate, so today we poll the Temporal Visibility API from an external service to detect failures.

Describe the behaviour you'd like

A worker- or workflow-level hook that fires on every workflow's terminal close, registered once (not per-workflow YAML, not per-activity). Conceptually an onComplete / onFailure callback or interceptor that receives:

• the workflow input and/or custom search attributes,
• the terminal status (succeeded / failed / timed_out / terminated / canceled),
• for failures, the error message / reason.

So a single registered handler can react to outcomes for all workflows.

Describe alternatives you've considered

• Per-activity try/catch + status step — works, but it's boilerplate in every activity of every workflow.
• use.catalog reusable blocks (#316) — reduces duplication, but wrapping every activity still means referencing the block per activity, so the boilerplate remains.
• CloudEvents (docs/dsl/debugging) — observability only: there's workflow.completed but no terminal workflow.faulted, the payload doesn't carry workflow input / search attributes / failure reason, and sinks are file/HTTP only.
• Nexus (#141) — on the roadmap, not available today.
• External Temporal Visibility polling — what we do now; it works, but it's a separate poller with its own checkpoint/dedup concerns and can't see in-workflow failure detail without extra history fetches.

Additional context

This is a common need (status reporting / alerting / external-system sync) that currently forces either per-activity boilerplate or an out-of-band watcher. A generic terminal-outcome hook would let us capture outcomes in-workflow and retire the external poller.

One concrete option, if a full hook is heavy: extend the existing CloudEvents to emit a terminal workflow.faulted event carrying input/search-attributes + failure reason to a consumable sink.

Happy to contribute a PR once the shape is agreed.

Describe the behaviour you'd like

A worker- or workflow-level hook that fires on every workflow's terminal close, registered once (not per-workflow YAML, not per-activity). Conceptually an onComplete / onFailure callback or interceptor that receives:

• the workflow input and/or custom search attributes,
• the terminal status (succeeded / failed / timed_out / terminated / canceled),
• for failures, the error message / reason.

So a single registered handler can react to outcomes for all workflows.