test(wake): pin tflite runtime in artifact-selection test; merge tflite_ok into available gate

Follow-up to benbarclay's macOS ARM64 tflite fix:
- test_openwakeword_bundled_model_matches_framework stubs
  ensure_tflite_runtime()=True so it exercises artifact selection, not
  runtime availability — off-Darwin the bridge legitimately returns
  False and the engine falls back to onnx (that path is covered by its
  own tests). Was failing on Linux CI otherwise.
- check_wake_word_requirements now ANDs both this branch's STT/TTS gate
  and Ben's tflite_ok into 'available' (cherry-pick conflict resolution).

tests/tools/test_wake_word.py

@@ -305,6 +305,10 @@ def test_openwakeword_default_resolves_to_bundled_model(monkeypatch, model_value
 
 def test_openwakeword_bundled_model_matches_framework(monkeypatch):
     calls = _install_fake_openwakeword(monkeypatch)
+    # Pin the tflite runtime as present so this exercises artifact selection,
+    # not runtime availability — off-Darwin the bridge legitimately returns
+    # False and the engine falls back to onnx (covered separately below).
+    monkeypatch.setattr(ww, "ensure_tflite_runtime", lambda: True)
     ww._OpenWakeWordEngine(
         {"provider": "openwakeword", "openwakeword": {"inference_framework": "tflite"}}
     )