feat: add call history API endpoints and TTS service client

Adds read-only access to persisted call records for the dashboard
and implements a client for the Rhema text-to-speech service.

- api/call_history.py: New router providing paged call lists
  and detailed call records with transcript metadata.
- services/tts.py: Async client for OpenAI-compatible TTS
  endpoints (Rhema/Kokoro) used for call-flow steps.

tests/test_tts.py

@@ -0,0 +1,72 @@
+"""Tests for the Rhema TTS client (services/tts.py)."""
+
+from unittest.mock import AsyncMock, MagicMock, patch
+
+import pytest
+
+from config import TTSSettings
+from services.tts import TTSService
+
+
+def _settings(**overrides) -> TTSSettings:
+    defaults = {
+        "base_url": "http://localhost:9000",
+        "model": "speaches-ai/Kokoro-82M-v1.0-ONNX",
+        "voice": "af_heart",
+        "api_key": "",
+        "timeout": 5.0,
+    }
+    defaults.update(overrides)
+    return TTSSettings(**defaults)
+
+
+class TestTTSService:
+    @pytest.mark.asyncio
+    async def test_synthesize_sends_expected_body(self):
+        svc = TTSService(_settings())
+
+        mock_response = MagicMock()
+        mock_response.content = b"RIFFfake-wav-bytes"
+        mock_response.raise_for_status = MagicMock()
+
+        mock_client = MagicMock()
+        mock_client.is_closed = False
+        mock_client.post = AsyncMock(return_value=mock_response)
+
+        with patch.object(svc, "_get_client", AsyncMock(return_value=mock_client)):
+            audio = await svc.synthesize("hello world", voice="am_michael")
+
+        assert audio == b"RIFFfake-wav-bytes"
+        mock_client.post.assert_awaited_once()
+        call_args = mock_client.post.call_args
+        assert call_args.args[0] == "/v1/audio/speech"
+        body = call_args.kwargs["json"]
+        assert body["input"] == "hello world"
+        assert body["voice"] == "am_michael"
+        assert body["response_format"] == "wav"
+
+    @pytest.mark.asyncio
+    async def test_synthesize_empty_text_returns_empty(self):
+        svc = TTSService(_settings())
+        assert await svc.synthesize("") == b""
+        assert await svc.synthesize("   ") == b""
+
+    @pytest.mark.asyncio
+    async def test_synthesize_to_file_writes_audio(self, tmp_path):
+        svc = TTSService(_settings())
+        target = tmp_path / "out.wav"
+
+        with patch.object(svc, "synthesize", AsyncMock(return_value=b"WAVE-bytes")):
+            ok = await svc.synthesize_to_file("hello", str(target))
+
+        assert ok is True
+        assert target.read_bytes() == b"WAVE-bytes"
+
+    @pytest.mark.asyncio
+    async def test_bearer_header_used_when_api_key_set(self):
+        svc = TTSService(_settings(api_key="secret-token"))
+        client = await svc._get_client()
+        try:
+            assert client.headers.get("authorization") == "Bearer secret-token"
+        finally:
+            await svc.close()