Add subsquid host metrics

[dewabisma]

Previously we only get the processor metric now we also add the host metric

[n13]

Reviewer model: GPT 5.6 Sol

Verdict: REQUEST_CHANGES — the new host scrape is wired into the selector and metric panels, but the proc-1 status tile still monitors the old processor scrape.

• [P1] Point the proc-1 host-status tile at the new host-metrics job (grafana/dashboards/infrastructure/subsquid-host.json:36). The PR maps proc-1 to ${fleet}-subsquid-proc-1-hm for the node-exporter panels, while this tile continues to query ${fleet}-subsquid-proc-1, which is the processor metrics endpoint. If node_exporter or its route fails while the processor endpoint remains available, the Subsquid Host dashboard reports proc-1 as UP even though all of its host metrics are unavailable. Change this query to ${fleet}-subsquid-proc-1-hm, or expose processor and host-exporter health as distinct tiles if both signals are intentional.

Validation: reviewed the complete diff from 14e0d019c33acaa4ba11b39626115e774e815bb2 to 98e250621416e684ee68b4ac37b848927b926aa4; Prometheus 3.14.0 promtool check config passed; YAML parsing confirmed 36 unique scrape jobs and both new host jobs; dashboard JSON parsing, targeted Prettier, and git diff --check passed. Both new DNS names resolve and return the expected unauthenticated HTTP 401 from Cloudflare Access. The repository-wide formatting check remains limited by three unchanged files.

[n13]

Reviewer model: GPT 5.6 Sol

Verdict: APPROVE — the new Subsquid processor-host node_exporter scrape is correctly wired into Prometheus and the host dashboard, including the prior status-tile fix.

No blocking findings. Processor application metrics remain on ${fleet}-subsquid-proc-1, while host metrics and the proc-1 selector/status tile consistently use ${fleet}-subsquid-proc-1-hm for both Planck and mainnet.

Validation:

• Reviewed the complete diff from 14e0d019c33acaa4ba11b39626115e774e815bb2 to f24607ccbc14e86b87d86988938c6ee5a3607878 and the follow-up fix commit.
• Prometheus 3.14.0 promtool check config passed using a SHA-256-verified official release binary.
• Dashboard JSON parsing, semantic YAML checks (36 unique jobs and both expected targets/Cloudflare headers), targeted Prettier, and git diff --check passed.
• Both new hostnames resolve and return the expected unauthenticated HTTP 401 from Cloudflare Access.
• No GitHub checks are configured for this PR.