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/agent-transcribe.test.ts

@@ -0,0 +1,23 @@
+import assert from "node:assert/strict";
+import { describe, it } from "node:test";
+import { transcribeAudioFile } from "../../src/lib/llm/transcribe";
+
+describe("transcribeAudioFile", () => {
+  it("returns mock text when llm base url is unset", async () => {
+    const original = process.env.LLM_BASE_URL;
+    const originalProvider = process.env.LLM_PROVIDER;
+    delete process.env.LLM_BASE_URL;
+    delete process.env.LLM_PROVIDER;
+
+    const text = await transcribeAudioFile(
+      new Blob(["audio"], { type: "audio/wav" }),
+      "recording.wav",
+    );
+    assert.match(text, /milk/i);
+
+    if (original === undefined) delete process.env.LLM_BASE_URL;
+    else process.env.LLM_BASE_URL = original;
+    if (originalProvider === undefined) delete process.env.LLM_PROVIDER;
+    else process.env.LLM_PROVIDER = originalProvider;
+  });
+});