Harden update-bindings PR publishing against transient GitHub 502s

[Copilot]

The update-bindings / publish-pr job was failing intermittently when gh pr create -f hit a transient GitHub HTTP 502, causing the entire workflow to fail despite valid generated changes. This update makes PR publication resilient to short-lived API failures.

• Workflow reliability: retry PR creation in publish-pr

  • Updated .github/workflows/update-bindings.yml in the Publish PR step.
  • Wrapped gh pr create -f in bounded retry logic (3 attempts) with incremental backoff.
  • Preserves fail-fast behavior after the final attempt to avoid masking persistent errors.

• Scope

  • No changes to binding generation, artifact merge, or branch update logic.
  • Change is isolated to PR creation error handling.

for attempt in 1 2 3; do
  if gh pr create -f; then
    break
  fi

  if [ "${attempt}" -eq 3 ]; then
    echo "Failed to publish PR after ${attempt} attempts"
    exit 1
  fi

  echo "gh pr create failed, retrying..."
  sleep $((attempt * 5))
done

[wravery]

I'm not sure this will fix it, it seems to have hit a similar error the last 3 times it ran this workflow. This is worth a try, though.