wf: faster list, IDs-only output, and quieter port-forward teardown

[JoshVanL]

• workflow list: fetch per-instance metadata concurrently via errgroup (cap 32 in-flight) instead of serially, which was the bottleneck for apps with many instances.
• workflow list: add -o ids output, which skips the metadata fetch entirely and prints one instance ID per line. Errors out if combined with filters since those require metadata.
• portforward: filter out the noisy klog "Unhandled Error" lines emitted
  by client-go during teardown for benign "broken pipe", "connection reset by peer", and "use of closed network connection" cases, while letting all other errors fall through to the default handlers.

[Copilot]

Pull request overview

This PR improves workflow instance listing performance and ergonomics, and reduces noisy Kubernetes port-forward teardown logs in the CLI.

Changes:

• Speed up workflow list by fetching per-instance workflow metadata concurrently (errgroup with a concurrency cap).
• Add workflow list -o ids to print instance IDs only (skipping metadata fetch) and reject filters that require metadata.
• Filter out known-benign “Unhandled Error” klog lines emitted by client-go during port-forward teardown, while preserving default handling for other errors.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
pkg/workflow/list.go	Adds ListIDs, introduces concurrent metadata fetch with an errgroup limit, and keeps output sorted.
pkg/kubernetes/portforward.go	Wraps apimachinery global error handlers to suppress known-benign teardown error strings.
cmd/workflow/workflow.go	Extends --output validation/usage to support additional formats (including ids).
cmd/workflow/list.go	Implements -o ids output path and blocks combining it with metadata-dependent filters.
go.mod	Promotes golang.org/x/sync to a direct dependency for errgroup usage.

Comments suppressed due to low confidence (1)

pkg/workflow/list.go:234

• Because listOutput is built concurrently, its initial order is nondeterministic. The subsequent sort.SliceStable only compares Created, so workflows with identical Created timestamps will appear in a random order run-to-run. Add a deterministic tie-breaker (e.g., compare InstanceID when Created is equal) or preserve the original index when collecting results.

	sort.SliceStable(listOutput, func(i, j int) bool {
		if listOutput[i].Created.IsZero() {
			return false
		}
		if listOutput[j].Created.IsZero() {

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[acroca]

Is it the default order by creation time? I think I'd keep it in whatever order we receive, the user can always sort if they need it sorted for any reason.