docs: document exporter.staleness_threshold in README

[softberries]

Summary

• Adds a commented example of staleness_threshold to the main [exporter] config block in README.md.
• Adds a paragraph under "Recommended Settings by Cluster Size" explaining that the default auto-scales with poll_interval (×3) — important context for operators raising poll interval to 60–180s on large clusters.

Follow-up documentation for #76 (code, merged) and #77 (helm chart, open).

Why

Operators landing in the large-cluster section to bump poll_interval previously had no signal about how the staleness threshold behaves — they might assume it's still 90s and see metrics disappear between cycles. Now they know the default scales automatically and how to override it.

Test plan

• Markdown renders correctly (visual check).
• Reviewer sanity-check: the phrasing matches the actual code behaviour (poll_interval * 3 fallback, Option<Duration> override in config).

[Copilot]

Pull request overview

This PR updates the operator documentation to clarify how the exporter’s metrics staleness filtering behaves, specifically introducing and explaining exporter.staleness_threshold and its default behavior as poll_interval * 3.

Changes:

• Adds a commented staleness_threshold example to the main [exporter] config snippet in README.md.
• Adds guidance in the large-cluster tuning section explaining that the default staleness threshold scales with poll_interval (×3) and when to override it.

---

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