feat(agent): add voice input and photo attachments to assistant

Wire mic through Whisper-compatible transcriptions on LLM_BASE_URL.

Photos upload to MinIO and reach the vision model as base64 image_url parts.

tests/unit/assistant-chat-storage.test.ts

@@ -4,6 +4,7 @@ import {
   clearAssistantChat,
   loadAssistantChat,
   saveAssistantChat,
+  toClientChatMessage,
 } from "../../src/modules/agent/assistant-chat-storage";
 
 const storage = new Map<string, string>();
@@ -49,4 +50,15 @@ describe("assistant chat storage", () => {
     clearAssistantChat("user-a");
     assert.deepEqual(loadAssistantChat("user-a"), []);
   });
+
+  it("maps stored image messages to client attachments", () => {
+    const message = toClientChatMessage({
+      role: "user",
+      content: "read this",
+      imageUrl: "/api/uploads/assistant/home/photo.jpg",
+    });
+    assert.deepEqual(message.attachments, [
+      { type: "image", url: "/api/uploads/assistant/home/photo.jpg" },
+    ]);
+  });
 });