docs(k8s-reporter): lead Karpenter guidance with pinning, widening as last resort

[AlexKantor87]

What

Reorders the "Running on EKS with Karpenter" guidance in the k8s-reporter chart docs so pinning is the recommended fix and widening the report interval is a caveated last resort. Follow-up to #985.

Why

Frequent snapshots are how Kosli surfaces drift or an unauthorized change quickly. Widening the reporter interval to placate Karpenter trades that detection speed away, so it should not be the headline recommendation. Pinning the reporter to a stable node group (or running it out-of-cluster) solves the consolidation interference while keeping the 5-minute cadence.

Changes

• Reorder the section in both _templates.gotmpl (chart README) and _mintlify_templates.gotmpl (docs.kosli.com source): 1) pin to a stable node group (recommended), 2) run out-of-cluster, 3) widen the interval (last resort, with the detection-speed caveat).
• Regenerate README.md.
• Bump chart 2.3.0 → 2.3.1 (docs only; no change to rendered manifests).

No Kubernetes manifest changes; helm lint passes.

Refs #984

🤖 Generated with Claude Code

[claude]

Claude finished @AlexKantor87's task in 40s —— View job

---

Review: docs(k8s-reporter) Karpenter guidance reorder

• Read PR diff and changed files
• Verify README is consistent with both templates
• Check for stale cross-references / section-number links
• Assess static-analysis relevance (Go build/vet)
• Post review

---

This is a docs-only PR — no Go source, manifests, or values.yaml changed. go build/go vet/go mod tidy aren't relevant here, so I skipped Phase 1 static analysis (nothing they'd touch).

What the PR does well ✅

• Sound rationale. Leading with pinning and demoting widening the interval to a caveated last resort is the right call — frequent snapshots are the whole point of the reporter, and the PR keeps that reasoning front-and-center in the new intro paragraph.
• Consistency across all three sources. The reorder is applied identically to _templates.gotmpl (chart README source), _mintlify_templates.gotmpl (docs.kosli.com source), and the regenerated README.md. I grepped for the moved headings and intro line — all three files match, section numbering is 1 → 2 → 3 with no gaps, and no stale ### 1/2/3 links exist elsewhere.
• Correct SemVer bump. 2.3.0 → 2.3.1 (PATCH) is appropriate for a docs-only change with no manifest impact, and the version badge in README.md was updated to match.
• The do-not-disrupt warning is preserved at the end of the section, unchanged — good, since that caveat is still valid regardless of ordering.

Minor observations (non-blocking)

1. "out of cluster" vs "out-of-cluster". In the new section-3 sentence "Only if you cannot pin the reporter or move it out of cluster:" — the adjacent heading/section uses "out of the cluster". Reads fine, just a slight phrasing inconsistency. Optional.
2. Slightly long intro sentence. The new paragraph (Frequent snapshots are what let Kosli surface drift... should be a last resort.) packs two ideas into one sentence. Splitting into two would read a touch easier, but it's clear as-is. Optional.
3. Regeneration provenance. The PR says README.md was regenerated. Since it's hand-verifiable here and matches _templates.gotmpl exactly, no concern — just worth confirming it came from the generator (helm-docs) rather than a manual edit, so future regenerations don't drift.

Verdict

Clean, well-reasoned documentation improvement. Content is consistent across all three sources, numbering and version bump are correct, and no broken references. Looks good to merge — the observations above are purely optional polish.
· chart/k8s-reporter-karpenter-docs-order