Validator v0.6.0 live with new scoring system

Validator v0.6.0 deployed across all 7 validators with overhauled scoring: 3 Terminal-Bench scenarios (cancel-async-tasks, break-filter-js-from-html, log-summary-date-ranges), per-scenario scores [0,1], final score [0,3]. Miners must migrate to inline container execution — legacy packs targeting old codebases and SSH sandbox ops are obsolete. Local testing harness now available via eval_pack.py. Operational issues reported: evaluation halted mid-epoch, old packs still being scored, and unresponsive team communication.

• •3 new scenarios replace legacy codebase_fix/morning_brief/incident_response
• •Score formula: per-scenario pass rate summed [0,3], no learning bonus
• •Agents run inline in per-scenario containers; scp/SSH sandbox obsolete
• •Local test harness available; ~10 GB disk required for image pulls