Make octave-reading source attribution bilingual

The heading above each Matins octave reading was a bare Latin string
(e.g. "Sermo sancti Bernárdi Abbátis") rendered as-is in both the
Latin and English columns, since the app shows both languages side
by side rather than toggling. Widen OctaveReadingText.source and the
lesson part's label to carry per-language text, and render it as its
own lang-columns row like the body text beneath it.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01L3wxvz3mPXiHxkPB5JpnmD

tests/hours/prime.test.ts

@@ -90,7 +90,9 @@ describe('resolveOrdo("prime", ...)', () => {
   it('resolves a Regula reading for the day, via the canonical-date table', () => {
     const ordo = resolveOrdo('prime', '2026-08-25');
     // 08-25 maps to canonical 04-25 (Rule ch. 67) in data/regula/table.yml.
-    const rule = ordo.parts.find((p) => p.kind === 'lesson' && p.label?.includes('67'));
+    const rule = ordo.parts.find(
+      (p) => p.kind === 'lesson' && typeof p.label === 'string' && p.label.includes('67'),
+    );
     expect(rule).toBeDefined();
   });