Connect reading plans to study/audio, cache offline audio, add journal export

Reading plan days now offer "Study this" (seeds a new chunk-builder project
with the day's full passage and jumps straight into the Study editor) and
"Listen" (starts book-mode audio at the right chapter via a new
handlePlayBookAudioFrom, avoiding a stale-closure bug the reader-state
approach would hit). Fixed the "Listen to BSB Audio" player only rendering
when a study project existed, leaving Listen with nowhere to show up.

The service worker now caches audio.bible.helloao.org (the actual host
serving narrator audio, not just the text/cross-ref API) and no longer
tries to cache 206 Partial Content responses, which the Cache API rejects
and was throwing unhandled rejections for every range-requested audio file.

Added a reading journal export (Settings > My reading journal): gathers
bookmarks, text highlights, and completed plans, fetches their verse text,
and offers a .docx download or a print-to-PDF page.

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

public/sw.js

@@ -1,6 +1,9 @@
 const SHELL = 'bible-shell-v1';
 const API   = 'bible-api-v1';
-const BIBLE_ORIGINS = ['https://bible.helloao.org', 'https://bolls.life'];
+// Chapter text, interlinear cross-refs, and (BSB) chapter audio all come from
+// these origins — cache them so the reader, reading plans, and audio all keep
+// working offline once you've visited a passage before.
+const BIBLE_ORIGINS = ['https://bible.helloao.org', 'https://bolls.life', 'https://audio.bible.helloao.org'];
 
 self.addEventListener('install', (e) => {
   e.waitUntil(
@@ -27,12 +30,14 @@ self.addEventListener('fetch', (e) => {
   // Never intercept API routes — let Express handle them
   if (new URL(url).pathname.startsWith('/api/')) return;
 
-  // External Bible API: network-first, fall back to cache
+  // External Bible API (text, cross-refs, audio): network-first, fall back to cache.
+  // Audio elements often issue Range requests, which come back as 206 Partial
+  // Content — the Cache API throws on those, so only cache full (200) responses.
   if (BIBLE_ORIGINS.some(o => url.startsWith(o))) {
     e.respondWith(
       fetch(request)
         .then(res => {
-          if (res.ok) caches.open(API).then(c => c.put(request, res.clone()));
+          if (res.status === 200) caches.open(API).then(c => c.put(request, res.clone()));
           return res;
         })
         .catch(() =>