[aleckhoury]

⚠️ This PR is a rename

It renames every criterion and check ID in two experimental rubric areas, artifact, configuration, and provenance and task contract and scoring authority. Scoring rules, anchors, and checks are unchanged. Only the IDs and the way reference docs point at entries change.

Stacked on #199. Review that first; this PR's base switches to main when #199 merges.

Why

Requested in #199 review: the old IDs came from the source rubric's workflow numbering.

• Criterion IDs didn't fit. CREATE-03 and skill_id: first_eval_create sit on criteria that apply to every task, whoever created it.
• Check IDs didn't say what they were. C01 was the judge check for one criterion and C02 for another, so the instructions had to list checks one by one.
• Reference docs linked by position (#/items/0), so reordering the JSON would silently break them.

What changes

• Criterion IDs name their area: task-contract.scoring-authority, provenance.grading-path.
• Check IDs add their method: .deterministic, .human-review, .llm-judge, .probe, with -1, -2 only when an item has more than one check of that method.
• Source IDs are kept: each criterion has source: {id, skill_id} and each check has source_id, so earlier results and the planning docs still map. Result rows carry source; skill_id is no longer filled from the source rubric.
• Docs link by ID: each reference section says which entry it describes instead of using a JSON position.
• The task contract scoring rules got shorter now that check IDs carry the method ("the .llm-judge checks" instead of three listed IDs).

Criterion ID map

Old	New
DISC-01-A	provenance.task-artifact-access
DISC-01-B	provenance.grading-path
DISC-01-C	provenance.grading-dependency-access
DISC-02-A	provenance.declared-configuration
DISC-02-B	provenance.observed-configuration
DISC-03-A	provenance.runtime-route
AUD-02	provenance.comparison-fairness
ENV-01	provenance.solver-tool-access
ENV-02-A	provenance.dependency-substitution
ENV-03	provenance.external-state
CREATE-07	provenance.evidence-revision-consistency
CREATE-03	task-contract.clear-contract
CREATE-04	task-contract.discoverable-requirements
CREATE-05-A	task-contract.scoring-authority

Check IDs follow the same pattern, for example CREATE-05-A-C02 → task-contract.scoring-authority.llm-judge and DISC-01-B-C03 → provenance.grading-path.human-review-1. The full map is in the source_id fields.

Impact

• Breaking for retained experimental outputs: rubric demo and measurement outputs now use the new IDs. As with earlier rubric renames, rebuild old outputs; no aliases are kept (docs/results.md).
• Mostly mechanical: of about 1,040 changed lines, ~430 are tests and ~320 are rubric JSON (the two definition files, panel expectations, and skill eval cases). Code changes are ID string swaps, plus one place that parsed a check ID suffix (endswith("C01")) and now compares the full ID.

Validation

• CI: this tree is identical to 68d7b90, which passed CI on feat(rubric): assess task contracts and scoring authority #199 before the rename moved here.
• Tests: uv run --locked python -m pytest -q -n auto tests/test_rubric_*.py tests/test_skill_contract.py tests/test_live_skill_evaluations.py tests/test_audit_*.py tests/test_reference_ownership.py tests/test_task_evidence.py tests/test_trace_loading.py reports 2725 passed.
• Demos: tools/run_rubric_measurements.py and tools/run_rubric_provenance_demo.py pass with unchanged outcomes under the new IDs.
• Structure tests: each check ID must name its item and method, source IDs must be unique, and reference docs may not use positional links.

Refs ASE-1376.