fix(reporting): distinguish unselected mandatory minimums

[rudycelekli]

A scoped run can leave B01/B08/P01 unselected while mandatory_minimums_passed remains true for the selected population. The interactive artifact currently renders that boolean as “Mandatory minimums PASS”, even though the native mandatory-minimum checker records these gates in mandatory_minimums_not_run. Markdown already distinguishes them.

Derive the interactive summary label from the recorded gate state: FAIL takes precedence, then NOT RUN, INCONCLUSIVE, and PASS. Render each unselected gate as NOT RUN. Retain the underlying gate boolean and scoring behavior.

Validation:

• Five regressions fail on unchanged main; 21 reporting/UTC controls pass after the fix, including a failed selected gate alongside an unselected gate.
• Native check_mandatory_minimums([], selected_ids={"B19"}) feeds the production payload and JavaScript renderer. Its header changes from PASS to NOT RUN and lists the three unevaluated gates as NOT RUN.
• Required Ruff 0.16.9, Bandit, layout validation and all eleven shipped example fixtures pass.
• Advisory scoped Mypy retains seven identical current-main scorecard diagnostics; the artifact module has none. Hosted CI is reported separately.

Controlled export of the native scoped-gate result, not a live model score:

{"selected_ids":["B19"],"mandatory_minimums_passed":true,"mandatory_minimums_not_run":["B01","B08","P01"],"interactive_label":"NOT RUN"}

[n-papaioannou]

Heads up Rudy, after #201 merged this clashed in test_artifact_diff_unscored.py (both PRs added tests at the end), so we merged main into your branch and kept both sets. Tests and CI pass. Shout if anything looks off.