Metric graphs: indicate when results come from a lower-fidelity / downsampled query tier

Problem Statement

Metric time-series charts can drop to a lower-fidelity query tier (for example, when a window exceeds the row cap scanned before downsampling). This introduces variation/dips that are artifacts of the query tier rather than real signal. Today nothing on the graph indicates that a lower-fidelity tier is in use, so users interpret these artifacts as real data.

Why the current behavior is insufficient
Users cannot distinguish fidelity/sampling artifacts from genuine metric behavior. This produces repeated confusion and support questions, and undermines trust in metric values, even when the underlying alerting (which uses a shorter, full-fidelity window) is accurate.

Solution Brainstorm

Display an indicator (badge, shaded region, or legend note) on the chart when results are served from a downsampled/lower-fidelity tier, with a tooltip explaining the tier and the affected time range. Optionally surface the threshold (e.g., row cap) and which ranges are full vs. reduced fidelity.

Product Area

Alerts

No files, tests, or entry points are named. Start by tracing how metric chart results expose query-tier or fidelity information, then determine how the chart can receive the affected ranges; done means reduced-fidelity results are visibly indicated with explanatory context without implying genuine metric changes.