feat: add voice audio reader for singleplayer tossups

[krishaygarg]

1. Summary / Objective

Adds an optional voice audio reader to singleplayer tossup practice. This allows users to listen to questions read aloud in a high-quality neural voice instead of only reading them visually.

2. Key Changes

• On-Demand Neural Voices: Added a route to synthesize Edge neural TTS on-the-fly, returning base64 audio and word timestamps.
• Speed Scaling: Speech rate dynamically matches the selected "Reading Speed" setting.
• Toggle UI: Added an "Enable audio reader" toggle that takes effect on the next question.

3. How to Test / Verification

1. Run npm install and ensure your local server is running.
2. Navigate to http://localhost:3000/play/tossups/.
3. Check the "Enable audio reader" toggle.
4. Click "Start" / "Next" to hear the question.
5. Verify that pressing Buzz immediately stops the audio and opens the input box.
6. Verify that if you do not buzz, the 5-second countdown timer starts smoothly only after the last word is spoken.

4. Dependencies / Breaking Changes

• New Package: Added node-edge-tts to the server dependencies for voice synthesis.
• No Breaking Changes: All audio reader logic is encapsulated in new files (AudioReaderController.js and QuestionAudioPlayer.js).