fix: prevent short-lived pod loss on failed snapshot export

[peatey]

Summary

• change short-lived pod reads to two-phase semantics by returning a copy without clearing the buffer
• add explicit short-lived pod acknowledgment to clear the buffer only after successful export
• update exporter flow to acknowledge short-lived pods only when all emitters succeed
• add/update tests for buffer acknowledgment behavior and exporter success/failure handling

Validation

• go test ./pkg/emitter ./internal/mocks
• pkg/cluster envtest suite could not run in this environment because /usr/local/kubebuilder/bin/etcd is missing

[Copilot]

Pull request overview

This PR updates short-lived pod buffering semantics to avoid losing pods on failed snapshot exports by introducing an explicit acknowledgment step and adjusting exporter behavior/tests accordingly.

Changes:

• Change GetAllShortLivedPods to return a copy (no buffer clearing on read) and add AcknowledgeShortLivedPods to explicitly clear the buffer.
• Update exporter flow to acknowledge short-lived pods only after all emitters succeed.
• Add/update tests covering repeated reads, acknowledgment behavior, and exporter success/failure handling.

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
pkg/emitter/exporter.go	Adds “ack after successful emission” logic to exporter loop.
pkg/emitter/exporter_test.go	Adds tests asserting acknowledgment happens only on successful emission.
pkg/cluster/dynamic.go	Implements two-phase short-lived pod semantics (copy on read + explicit acknowledge).
pkg/cluster/dynamic_test.go	Updates envtest to validate buffer is not cleared until acknowledgment.
pkg/cluster/clustercache.go	Extends ClusterCache interface with AcknowledgeShortLivedPods.
internal/mocks/mocks.go	Updates mock ClusterCache to implement AcknowledgeShortLivedPods.

---

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