feat(notifications): add Text-to-Speech (TTS) read aloud feature for incoming messages

[ralexandrec]

Summary

This PR introduces an optional Text-to-Speech (TTS) read-aloud feature for incoming mesh messages, designed to provide eyes-free / hands-free audio announcements on Wear OS smartwatches.

Key Highlights & Architecture

1. Text-to-Speech Integration (Notifier.kt):

   • Implements TextToSpeech.OnInitListener inside Notifier to synthesize incoming chat messages ("$senderName: ${message.text}").
   • Configures AudioAttributes (USAGE_ASSISTANT, CONTENT_TYPE_SPEECH) for proper Wear OS audio routing to speakers or Bluetooth headsets.
   • Leverages native Android offline TTS voice data for off-grid operation.

2. Power & Screen-Off Execution (Notifier.kt):

   • Acquires a brief PARTIAL_WAKE_LOCK (5s) upon message arrival to prevent CPU sleep during speech synthesis, allowing incoming messages to be announced even when the display is off.

3. User Control & Persistence (Prefs.kt, SettingsScreen.kt):

   • Adds a read_notifications_tts preference in Prefs (backed by SharedPreferences and StateFlow).
   • Adds a Wear Compose SwitchButton under the Messaging subheader in SettingsScreen titled "Read aloud (TTS)" with clear status subtitles.
   • Defaults to enabled (true) for immediate out-of-the-box functionality.

4. Locale Formatting Fix in Unit Tests (MeshModels.kt):

   • Passes Locale.US to String.format in formatDistance to ensure unit test consistency across non-US system locales (e.g. preventing , decimal separator issues in pt_BR).

Verification

• ✅ All 33 unit tests pass (BUILD SUCCESSFUL).
• ✅ Verified on target Wear OS hardware.