feat: shared transcription cache to avoid duplicate Whisper API calls

Both services (nanoclaw + nanoclaw-codex) now share cache/transcriptions/
directory. When one service transcribes audio, the other reads from cache
instead of making a second Whisper API call. Uses .pending file to
coordinate concurrent transcription attempts.

src/channels/discord.test.ts

@@ -13,6 +13,7 @@ vi.mock('../config.js', () => ({
   ASSISTANT_NAME: 'Andy',
   TRIGGER_PATTERN: /^@Andy\b/i,
   DATA_DIR: '/tmp/nanoclaw-test-data',
+  CACHE_DIR: '/tmp/nanoclaw-test-cache',
 }));
 
 // Mock logger