test: add 157 functional tests covering 8 low-coverage modules

Analyze test coverage (75.3% → 85.4%) and add functional test suites
for the major gaps identified:

- test_agent_core.py: Full coverage for agent_core/interface.py (0→100%)
  and agent_core/ollama_adapter.py (0→100%) — data classes, factories,
  abstract enforcement, perceive/reason/act/recall workflow, effect logging

- test_docker_runner.py: Full coverage for swarm/docker_runner.py (0→100%)
  — container spawn/stop/list lifecycle with mocked subprocess

- test_timmy_tools.py: Tool usage tracking, persona toolkit mapping,
  catalog generation, graceful degradation without Agno

- test_routes_tools.py: /tools page, API stats endpoint, and WebSocket
  /swarm/live connect/disconnect/send lifecycle (41→82%)

- test_voice_tts_functional.py: VoiceTTS init, speak, volume clamping,
  voice listing, graceful degradation (41→94%)

- test_watchdog_functional.py: _run_tests, watch loop state transitions,
  regression detection, KeyboardInterrupt (47→97%)

- test_lnd_backend.py: LND init from params/env, grpc stub enforcement,
  method-level BackendNotAvailableError, settle returns False (25→61%)

- test_swarm_routes_functional.py: Agent spawn/stop, task CRUD, auction,
  insights, UI partials, error paths (63→92%)

https://claude.ai/code/session_01WU4h3cQQiouMwmgYmAgkMM

tests/test_watchdog_functional.py

@@ -0,0 +1,100 @@
+"""Functional tests for self_tdd.watchdog — continuous test runner.
+
+All subprocess calls are mocked to avoid running real pytest.
+"""
+
+from unittest.mock import patch, MagicMock, call
+
+import pytest
+
+from self_tdd.watchdog import _run_tests, watch
+
+
+class TestRunTests:
+    @patch("self_tdd.watchdog.subprocess.run")
+    def test_run_tests_passing(self, mock_run):
+        mock_run.return_value = MagicMock(
+            returncode=0,
+            stdout="5 passed\n",
+            stderr="",
+        )
+        passed, output = _run_tests()
+        assert passed is True
+        assert "5 passed" in output
+
+    @patch("self_tdd.watchdog.subprocess.run")
+    def test_run_tests_failing(self, mock_run):
+        mock_run.return_value = MagicMock(
+            returncode=1,
+            stdout="2 failed, 3 passed\n",
+            stderr="ERRORS",
+        )
+        passed, output = _run_tests()
+        assert passed is False
+        assert "2 failed" in output
+        assert "ERRORS" in output
+
+    @patch("self_tdd.watchdog.subprocess.run")
+    def test_run_tests_command_format(self, mock_run):
+        mock_run.return_value = MagicMock(returncode=0, stdout="", stderr="")
+        _run_tests()
+        cmd = mock_run.call_args[0][0]
+        assert "pytest" in " ".join(cmd)
+        assert "tests/" in cmd
+        assert "-q" in cmd
+        assert "--tb=short" in cmd
+        assert mock_run.call_args[1]["capture_output"] is True
+        assert mock_run.call_args[1]["text"] is True
+
+
+class TestWatch:
+    @patch("self_tdd.watchdog.time.sleep")
+    @patch("self_tdd.watchdog._run_tests")
+    @patch("self_tdd.watchdog.typer")
+    def test_watch_first_pass(self, mock_typer, mock_tests, mock_sleep):
+        """First iteration: None→passing → should print green message."""
+        call_count = 0
+
+        def side_effect():
+            nonlocal call_count
+            call_count += 1
+            if call_count >= 2:
+                raise KeyboardInterrupt
+            return (True, "all good")
+
+        mock_tests.side_effect = side_effect
+        watch(interval=10)
+        # Should have printed green "All tests passing" message
+        mock_typer.secho.assert_called()
+
+    @patch("self_tdd.watchdog.time.sleep")
+    @patch("self_tdd.watchdog._run_tests")
+    @patch("self_tdd.watchdog.typer")
+    def test_watch_regression(self, mock_typer, mock_tests, mock_sleep):
+        """Regression: passing→failing → should print red message + output."""
+        results = [(True, "ok"), (False, "FAILED: test_foo"), KeyboardInterrupt]
+        idx = 0
+
+        def side_effect():
+            nonlocal idx
+            if idx >= len(results):
+                raise KeyboardInterrupt
+            r = results[idx]
+            idx += 1
+            if isinstance(r, type) and issubclass(r, BaseException):
+                raise r()
+            return r
+
+        mock_tests.side_effect = side_effect
+        watch(interval=5)
+        # Should have printed red "Regression detected" at some point
+        secho_calls = [str(c) for c in mock_typer.secho.call_args_list]
+        assert any("Regression" in c for c in secho_calls) or any("RED" in c for c in secho_calls)
+
+    @patch("self_tdd.watchdog.time.sleep")
+    @patch("self_tdd.watchdog._run_tests")
+    @patch("self_tdd.watchdog.typer")
+    def test_watch_keyboard_interrupt(self, mock_typer, mock_tests, mock_sleep):
+        mock_tests.side_effect = KeyboardInterrupt
+        watch(interval=60)
+        mock_typer.echo.assert_called()  # "Watchdog stopped"