Gate the outcome band against a note that runs out of its box

The outcome boxes sit at a fixed width and neither line inside one wraps
or clips, so an over-long note simply runs into the next box with nothing
failing. That was the one silent overflow the suite did not cover: card
notes are checked against the wrapper, and the tagline against its rule,
but the terminal band was unguarded.

The REFUTED note was one character over the budget the new test derives,
so it is reworded. Like the tagline budget it counts characters rather
than measuring glyphs, and it is anchored to the widest note that has
been read on a rendered page.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

Author: HarperZ9

docs/art/claim-oracle.svg

@@ -28,7 +28,7 @@
       ],
       "outcomes": [
         {"label": "MATCH", "note": "the graph carries the edge, with file and line", "tone": "verified"},
-        {"label": "REFUTED", "note": "the graph is complete here and has no such edge", "tone": "drift"},
+        {"label": "REFUTED", "note": "the graph is complete here, with no such edge", "tone": "drift"},
         {"label": "UNVERIFIABLE", "note": "an endpoint sits outside the scanned workspace", "tone": "none"}
       ]
     }

docs/art/index.art.json

@@ -152,3 +152,29 @@ def test_the_tagline_stays_inside_its_rule():
         tagline = spec["header"]["tagline"]
         assert len(tagline) <= TAGLINE_BUDGET, (
             f"{len(tagline)} characters runs past the rule: {tagline!r}")
+
+
+# An outcome box is one unwrapped line of label over one unwrapped line of note,
+# inset 14px inside a box that is (960 - 88 - 26 * (n - 1)) / n wide. Neither
+# line wraps and neither is clipped, so an over-long note simply runs out of its
+# box and into the next one. Like the tagline budget above, these count
+# characters rather than measure glyphs: a guardrail, not a typographic fact.
+# The widths come from the box at three outcomes, which is what every spec so
+# far uses, and shrink with the box when a spec uses more.
+def _outcome_budgets(count: int) -> tuple[int, int]:
+    span = (960 - 44 * 2 - 26 * (count - 1)) / count
+    usable = span - 14 - 10
+    return int(usable / 7.0), int(usable / 5.4)
+
+
+def test_no_outcome_runs_out_of_its_box():
+    for spec in _specs():
+        for flow in spec.get("flows", []):
+            outcomes = flow["outcomes"]
+            label_budget, note_budget = _outcome_budgets(len(outcomes))
+            for item in outcomes:
+                assert len(item["label"]) <= label_budget, (
+                    f'{item["label"]!r} is wider than its box')
+                assert len(item["note"]) <= note_budget, (
+                    f'the note under {item["label"]} is wider than its box: '
+                    f'{item["note"]!r}')