fix(foreman): make the CustomResourceState config produce usable metrics

[Defilan]

What

Make foreman.crs.enabled produce metrics — it has never emitted a single one —
and add the labels a fleet view actually needs.

Why

Fixes #1643. Refs #1602, #1640.

Four defects, each reproduced against a live kube-state-metrics, detailed in
#1643: the ConfigMap was not a complete CustomResourceStateMetrics document
(KSM silently parses an empty config), gauge.valueFrom used an object form
that crashloops KSM, valueFrom is relative to a required path the config
omitted (parses fine, emits nothing), and enum fields used type: Gauge with a
fabricated labelsMapping key instead of StateSet.

How

• Wrap the config as kind: CustomResourceStateMetrics / spec:.
• path + relative valueFrom for the numeric gauges; StateSet with the CRs'
  real enum values for phase/verdict.
• The shipped dashboard's queries move with the enum casing
  (phase="ready" → phase="Ready").
• Drops foreman_agentictask_tasks_failed: with phase a real StateSet it is
  exactly phase{phase="Failed"}, nothing queried it, and it cost 630 series.
• New labels so the fleet view can say what a node is and what it is doing:
  every resource stamps name; FleetNode info gains current_task, version,
  agent_kind, and kubernetes_node (the [BUG] FLEET_NODE_NAME is set from spec.nodeName but never read, so FleetNode identity is the pod, not the node #1640 property, shipped in feat(foreman): report the Kubernetes node on FleetNode.status, not as identity #1649);
  role_info is new, one series per role (spec.roles is a list — a joined
  label would make "which nodes can review?" a substring match); AgenticTask
  info gains node and kind, the join a fleet table is built on.

Deployment notes recorded in #1643: kube-prometheus-stack's explicit
--resources list excludes custom resources, and KSM needs RBAC for the
foreman API group — both outside what this chart can ship.

Verification

Deployed end-to-end on a live cluster: all 14 metric families emit
(foreman_fleetnode_phase 40 series, role_info 23, heartbeat 10, …) and every
query in the shipped dashboard returns data. helm unittest: 47/47, including
regression tests for the document wrapper, the valueFrom shape, and the
StateSet conversion.

Checklist

• Tests added/updated
• make test passes locally
• make lint passes locally
• Commit messages follow conventional commits
• All commits are signed off (git commit -s) per DCO
• AI assistance (if any) is disclosed above, per CONTRIBUTING.md
• Documentation updated (if user-facing change) — values.yaml comments cover the surface

Assisted-by: Claude Code (diagnosed the four defects against a live
kube-state-metrics with a three-way probe, implemented the fixes, and verified
every metric family and dashboard query live before submitting).

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!