[squad-plan] Remove redundant live CI re-check from Avenger Step 0 prompt

[github-actions]

Objective

Simplify Avenger's Step 0 so it uses the already-provided CI status outputs instead of making a duplicate gh run list call.

Issue context

Issue #50312 found Avenger already receives ${{ needs.check_ci_status.outputs.ci_status }} and ci_run_id from the check_ci_status job, but the prompt tells the agent to re-check live CI via gh run list in Step 0. That extra tool call adds avoidable tokens/cache-read volume and latency (~10-15% estimated AIC savings).

Implementation guidance

• Edit .github/workflows/avenger.md.
• Update Step 0 instructions to:
  • immediately noop if the passed-in CI status is success
  • proceed directly to the repair sequence if CI status is failure
• Remove the redundant live API re-verification block unless there is a strong documented reason to keep a smaller/leaner version (e.g., staleness check only when ambiguous).
• Keep the prompt concise and consistent with the workflow's existing job outputs.
• Run make recompile after editing so the .lock.yml is regenerated.

Acceptance criteria

• Avenger prompt no longer instructs the agent to run a redundant gh run list in Step 0 when CI status is already known.
• Step 0 clearly uses needs.check_ci_status.outputs.ci_status as the source of truth for initial gating.
• make recompile succeeds and the generated lock file updates cleanly with no unrelated diffs.

Dependency / ordering notes

Independent of the model-mismatch fix; can be implemented in parallel. Should be validated (recompiled) after any other Avenger markdown edits to avoid merge conflicts in the .lock.yml.

Generated by 🧑‍🤝‍🧑 Squad Plan · auto · 37.1 AIC · ⌖ 2.24 AIC · ⊞ 9.5K · ◷
_{Comment /squad-plan to run again}