v0.2: streaming playback, model params, text input

- Stream audio directly to device via BufferedWaveProvider (no temp WAVs)
- Add Noise/Speed/NoiseW sliders with live value labels
- Sliders reinit engine on MouseUp (not Scroll) to avoid locking
- Add text input field with Speak button for direct synthesis
- Test button uses text input if non-empty
- Fix sentence terminator auto-append (espeak-ng final-word drop)
- Fix UTF-8 text marshaling (piper_synthesize_start keeps text ref)
- Fix voice catalogue language column (JSON snake_case mapping)
- Use piper_default_synthesize_options for model-specific defaults
- Fix nullable warnings in designer files
- Fix unused _playing field in AudioOutput

Robovoice.App/AudioOutput.cs

@@ -8,7 +8,6 @@ internal sealed class AudioOutput : IDisposable
     private BufferedWaveProvider? _bufferProvider;
     private int _sampleRate;
     private string _deviceName = string.Empty;
-    private bool _playing;
     private bool _disposed;
 
     public Action<string>? Log { get; set; }
@@ -35,7 +34,6 @@ internal sealed class AudioOutput : IDisposable
 
         _waveOut.Init(_bufferProvider);
         _waveOut.PlaybackStopped += OnPlaybackStopped;
-        _playing = true;
         _waveOut.Play();
     }
 
@@ -60,7 +58,6 @@ internal sealed class AudioOutput : IDisposable
 
     private void OnPlaybackStopped(object? sender, StoppedEventArgs e)
     {
-        _playing = false;
     }
 
     public void Stop()
@@ -74,7 +71,6 @@ internal sealed class AudioOutput : IDisposable
         }
         _bufferProvider?.ClearBuffer();
         _bufferProvider = null;
-        _playing = false;
     }
 
     private static int FindDevice(string? deviceName)