[julianocosta89]

Description

First step of the "embed links" part of the exemplars roadmap (perses/perses#3445, step 2): extends the query-result contract with an optional links field on Exemplar.

The design keeps the division of responsibilities clean:

• The query plugin (PrometheusTimeSeriesQuery, follow-up PR) reads the datasource exemplars configuration (URL template + trace ID label) and computes the final URLs per exemplar, attaching them here.
• The panels and the shared tooltip stay datasource-agnostic: they just render ExemplarLink[] as clickable buttons. Any panel that displays exemplars gets link support for free, and any query plugin can produce links without the UI knowing anything about the datasource.

export interface ExemplarLink {
  /** Optional display label for the link (e.g. shown on the button in the tooltip). */
  name?: string;
  /** Final URL, with variables and trace ID placeholders already resolved. */
  url: string;
}

export interface Exemplar {
  labels: Labels;
  value: number;
  timestamp: UnixTimeMs;
  /**
   * Links related to this exemplar (e.g. to a tracing UI), computed by the query
   * plugin from the datasource exemplars configuration. Optional so plugins that
   * don't support links keep the payload unchanged.
   */
  links?: ExemplarLink[];
}

The field is optional, so the change is fully backwards compatible: older dashboards/datasources load unchanged, and plugins that don't produce links keep the payload identical.

Testing

• Type-check and the full existing spec test suite pass (14 tests).
• No CUE / Go / Java counterpart: TimeSeriesData and its exemplar types exist only in the TS part of the spec (query-result contract, not a persisted resource) — verified there are no exemplar references in cue/, go/ or java/.

[jgbernalp]

@julianocosta89 can you please sign your commits?

[Nexucis]

@julianocosta89 do you mind to signoff your commit, we will then be able to move forward

[julianocosta89]

@jgbernalp and @Nexucis sorry, I was off last week and couldn't do it.
I've just pushed the signed commit, sorry about that

[julianocosta89]

I believe the failing CI is not related to this PR

[jgbernalp]

We fixed the CI in #125, mind rebasing?