Does Dynatrace deduplicate writes when a custom plugin re-executes after a failure?

We run several custom OneAgent plugins that push measurements into our Dynatrace managed cluster. When a plugin execution fails partway through, the plugin is re-executed on the next cycle, and I want to confirm whether the retried execution can produce duplicate data points or events for the same timestamp.

For built-in paths I understand the platform handles this reasonably well: PurePath segments rely on request IDs to detect retried calls, the Events API treats payloads as idempotent when a stable eventId is supplied, and problem detection deduplicates by fingerprint. What I cannot find documented is whether custom plugin executions get any comparable deduplication, or whether the caller-side plugin code is fully responsible for idempotency, similar to direct API ingestion that bypasses OneAgent.

Our environment is a managed cluster, so SaaS-only behavior would not necessarily apply, and I would like to avoid relying on assumptions that vary across releases.

Does Dynatrace apply any server-side deduplication to measurements or events submitted by a retried custom plugin execution? If not, is supplying a deterministic timestamp or identifier per measurement the intended idempotency mechanism? Is there a documented way to verify duplicate suppression from the UI or API?