eval-magic v0.3.1

What's Changed

• Merge pull request #62 from slowdini/dev by @slowdini in #63
• feat(headless): add model selection support to headless run mode by @slowdini in #67
• fix(parity): add run capability matrix, to replace hard-coded codex ref by @slowdini in #68
• fix(promote-baseline): fix promote baseline for multi-run gradings by @slowdini in #71

Full Changelog: v0.3.0...v0.3.1