This PR implements high-impact changes to reduce GitHub Actions spend and improve issue backlog management.
#### 📉 Actions Spend Reductions
- **Mac Runner Downgrade**: Switched `test_mac` and `e2e_mac` jobs from `macos-latest-large` to standard `macos-latest`. Mac Large runners are significantly more expensive (~40x Linux) and were the primary driver of the recent 10x spend spike.
- **Mac Matrix Optimization**: Reduced the `test_mac` matrix to only Node 20.x. Testing 3 Node versions on Mac for every PR was redundant, as OS-specific issues are typically caught by the LTS version, while cross-version logic is verified on the much cheaper Linux runners.
- **Pulse Workflow Optimization**: Added a `precheck` job to the Pulse workflow to skip the expensive `npm ci` and setup steps if no reflex scripts are present.
#### 📋 Backlog Management
- **Accelerated Stale Lifecycle**: Reduced `STALE_DAYS` from 60 to 30. This ensures that inactive issues are identified and processed faster, helping to manage the current volume of 2148 open issues.
- **Increased Triage Throughput**: Increased the batch limit for scheduled issue triage from 100 to 200, allowing the bot to process a larger portion of the untriaged backlog in each run.
#### 📊 Expected Impact
- **Cost**: Significant reduction in monthly Actions spend by eliminating redundant Mac Large runner minutes.
- **Throughput**: Faster closure of stale issues and double the triage capacity per hour.
---
*This PR was generated by the Gemini CLI Bot (Brain Layer) based on time-series metric analysis.*
gemini-cli[bot]