v0.0.6 — 3200+ Stress Tests + CI
Changes from v0.0.5
- 1,800 randomised within-capacity trials (0–8 errors) — 100% recovery
- 1,400 randomised overcapacity trials (9–64 errors) — 0% miscorrection
- Throughput benchmark
- GitHub Actions CI (Node 20.x + 22.x)
- fix(solver): rank-deficient system handling