Energy metering and carbon footprint reporting

[jeremymanning]

Description

Per spec T144b and Constitution Principle IV, the system must track energy consumption and reduce joules-per-useful-result year over year.

Requirements

• Per-node energy metering: CPU-time, GPU-time, estimated watts (via RAPL on Intel, PowerCap on Linux, IOReport on macOS)
• Aggregate cluster carbon footprint calculation using regional carbon intensity data
• Reporting endpoint for cluster-wide energy and carbon metrics
• Published energy footprint per the whitepaper ("publishes its aggregate energy and carbon footprint")
• Year-over-year joules-per-useful-result tracking

Success Criteria

• Per-node energy estimates within ±20% of real power draw
• Cluster-wide energy aggregation across all participating nodes
• Carbon footprint calculation using regional grid intensity
• Reporting endpoint accessible via CLI and REST
• Integration test with real hardware power measurement comparison
• cargo test passes

Testing (Principle V)

• Compare RAPL readings against wall-meter power draw on test hardware
• Run workload → measure energy → verify report accuracy
• Multi-region test with different carbon intensities → verify correct aggregation

[jeremymanning]

Addressed in spec 004-full-implementation (PR #59). Implementation verified in code; 802 tests passing across Linux/macOS/Windows. See #57 master plan for full context.