v3.0.1-post1

Architecture and reliability update after the baseline release.
Highlights:

• Unified MedQA and HealthBench evaluation core across CLI and Workbench.
• Correct HealthBench per-sample records and fixed/scrollable result layout.
• Hardened credential handling, rubric/catalog validation, report ownership, auth, and production startup checks.
• Added cancellation-safe worker finalization, checkpointed artifacts, environment-backed credentials, worker leases, queue abstraction, and concurrent-claim coverage.
• Added full Python, mypy, frontend, and Playwright regression coverage.

v3.0.1 baseline

Baseline release for the remote main branch before the local medical-evals architecture and reliability updates.