feat(audio): add native audio support to type system and model clients

- Add AudioSource interface and 'audio' variant to MessageContentPart union
- Update buildUserMessage() to create audio content parts from attachments
- Add attachmentToAudioSource(), hasAudio(), stripAudioParts() helpers
- Gemini: native audio via inlineData (same format as images)
- OpenAI/GitHub: native audio via input_audio content parts
- Anthropic/Bedrock: graceful fallback to transcript text
- Update getMessageTextWithTools() to handle audio blocks for local models

src/models/gemini.ts

@@ -188,6 +188,15 @@ function convertMessages(messages: Message[]): Content[] {
         // so we pass as a text description. In production, you'd want to fetch + base64 encode.
         return { text: `[Image: ${part.source.url ?? 'unavailable'}]` };
       }
+      // Audio part — Gemini supports native audio via inlineData (same format as images)
+      if (part.type === 'audio') {
+        return {
+          inlineData: {
+            mimeType: part.source.media_type,
+            data: part.source.data,
+          },
+        };
+      }
       return { text: JSON.stringify(part) };
     });