docs: define candidate calibration protocols

[verlyn13]

What

• Split the calibration umbrella into candidate E4 grader-validation, E1 A/A invariance, and E2 operating-characteristics SAPs.
• Define E1 as an m=1 falsification protocol with fixed terminal counts and separate valid-negative, falsified, apparatus-failure, and invalid-evidence outcomes.
• Keep every SAP explicitly unregistered and observation-free.
• Make the exact local merge gate the CI entrypoint, single-source Python extras from pyproject.toml, and pin Quarto through .quarto-version.

Why

The active roadmap and calibration draft had drifted: the old E1 rank-correlation framing was not identified for one grader, candidate numbers were not consistently separated from observations, and CI duplicated a local gate that packaging metadata could silently diverge from.

Impact

This makes protocol review possible without authorizing data collection or changing any published scientific result. The frozen identity-domain specification and historical evidence artifacts are unchanged.

Verification

• python -m pip install '.[dev,report]' && bash scripts/run-merge-gate.sh
• Ruff lint and format checks
• 197 tests
• both exact-result checks
• temporary wheel containment build
• sample and campaign evidence validation
• Quarto site, sample-report, and campaign-report renders
• introduced-content publication-safety scan
• frozen identity-domain check

Pre-commit claim and public-boundary audit: ADMIT.

[cursor]

Bugbot is not enabled for your account, so this pull request was not reviewed.

Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs.