feat: add quersi monitoring

[dewabisma]

Summary

• Add prometheus scrape config
• Add grafana dashboard
• Add grafana alert rules

[n13]

Reviewed the full stack: scrape config, dashboard JSON, alert rule, routing, and README. Everything follows the established repo patterns and both YAML files parse cleanly with no duplicate rule UIDs or job names. No blocking issues.

What was checked:

• prometheus/prometheus.yml — the quersi-1 job is byte-for-byte the same CF-Access-protected scrape pattern as the existing senoti/subsquid jobs (same secrets files, protocols, timeouts). Omitting the chain label is correct for a chain-independent service.
• grafana/provisioning/alerting/rules.yml — quersi_host_down is an exact structural match of planck_senoti_host_down (reduce → threshold, noDataState: Alerting, for: 3m). Verified against policies.production.yml: with severity: critical and no chain label it falls through the chain-scoped routes to the generic "critical → Email" route, so notifications will fire as intended.
• grafana/dashboards/infrastructure/quersi-host.json — valid JSON, unique infra-quersi-host UID, unique panel IDs. Queries mirror support-host.json (the single-host template — correctly no job templating, unlike the fleet dashboards), all filtered on job="quersi-1" matching the new scrape job. The added Host Status stat is a nice touch the support dashboard lacks.
• README table entry matches the existing Infrastructure rows.

Minor (non-blocking): the job name quersi-1 sits between the two existing conventions (support-host/telemetry-host for single infra hosts vs planck-*-1 for chain fleets). Fine as is — just noting it since renaming later would break the dashboard and alert selectors.

Verdict: LGTM — approving.