ci: fix zizmor findings in GitHub Actions workflows

[danbarr]

Follow-up to the zizmor workflow added in the previous commit: addresses the findings it surfaces across the existing workflows now that the scan is in place.

• Sets persist-credentials explicitly on every checkout: false where the job never pushes, true (with a comment) where it does and genuinely needs the persisted credential.
• Disables setup-go's default caching in the container-build job that publishes runtime artifacts, per zizmor's cache-poisoning audit.
• Swaps the spoofable github.actor bot check in skill-version-check.yml for github.event.pull_request.user.login.
• Suppresses the dangerous-triggers finding on the two workflow_run based report workflows with a documented rationale: neither checks out or executes the triggering run's code, and the only untrusted value they consume (a PR number read from an artifact) only affects which PR gets commented on.

With the zizmor workflow gating on medium+ severity only, this clears everything at that threshold. The remaining low/informational findings (template-injection warnings on already-scoped env values,
one ad-hoc npm install) are left as-is.

Co-Authored-By: Claude Sonnet 5 noreply@anthropic.com

---

_{Stack created with GitHub Stacks CLI • Give Feedback 💬}

[toolhive-release-app]

🛡️ Skill Security Scan Results

✅ agent-observability-eval-bootstrap

• Status: Passed
• Findings: 4
• Allowed (not blocking): 1
  • MANIFEST_MISSING_LICENSE (Allowed: datadog-labs/agent-skills is licensed MIT at the repository root; upstream does not embed an SPDX license identifier in per-skill SKILL.md frontmatter.)

---

Summary: Scanned 1 skill(s), all passed security checks. ✅

[toolhive-release-app]

🔒 MCP Security Scan Results

✅ adb-mysql-mcp-server

• Status: Passed
• Tools scanned: 3
• Result: No security issues detected

✅ agentql-mcp

• Status: Passed
• Tools scanned: 1
• Result: No security issues detected

---

Summary: Scanned 2 MCP server(s), all passed security checks. ✅