Add Microsoft voice live service

[pragmatrix]

The goal of this PR is to add a basic service that supports Microsoft Voice Live API endpoints and to experiment with Semantic / Smart VAD.

[Copilot]

Pull request overview

Adds a new microsoft-voice-live service crate to support Microsoft Foundry Voice Live Realtime (transcription-focused) endpoints, and updates existing realtime event types to consistently serialize variant fields in camelCase (needed for OpenAI/Azure-style realtime JSON payloads). This also extends the transcribe example and docs to allow experimenting with Azure Semantic/Smart VAD parameters.

Changes:

• Introduce services/microsoft-voice-live (host/client/transcription buffering) and register it in the main registry + service re-exports.
• Normalize realtime Service{Input,Output}Event serde attributes to apply camelCase to enum variant fields consistently.
• Extend CLI example/docs/just targets to run the new Voice Live provider and provide env var configuration guidance.

Reviewed changes

Copilot reviewed 20 out of 21 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
src/lib.rs	Re-exports the new MicrosoftVoiceLiveTranscribe service from the top-level services module.
src/context_switch.rs	Registers the new service in the runtime service registry under microsoft-voice-live-transcribe.
services/openai-dialog/src/types.rs	Updates serde tagging/casing rules for realtime input/output event enums.
services/openai-dialog/src/transcription_state.rs	Adds new transcription buffering/state module for interim + final transcript assembly.
services/openai-dialog/src/lib.rs	Switches module wiring to use transcription_state.
services/openai-dialog/src/client.rs	Updates imports to the new transcription state module.
services/openai-dialog/Cargo.toml	Bumps tokio-tungstenite to 0.29.0.
services/microsoft-voice-live/src/lib.rs	New crate entrypoint exporting params/service/events.
services/microsoft-voice-live/src/transcribe.rs	Implements Service wrapper and defines service-level control events + defaults.
services/microsoft-voice-live/src/host.rs	Builds Voice Live websocket URL and connects via Azure realtime protocol settings.
services/microsoft-voice-live/src/client.rs	Implements websocket loop, session update, audio append, and transcription event handling.
services/microsoft-voice-live/src/transcription_state.rs	New minimal transcription buffer state (input-side only).
services/microsoft-voice-live/Cargo.toml	New crate dependencies and workspace integration.
services/google-dialog/src/types.rs	Aligns serde casing rules for realtime event enums.
README.md	Documents how to run the transcribe example with Voice Live + required env vars.
justfile	Adds a convenience target for Voice Live transcription in German.
examples/transcribe.rs	Adds voice-live provider option and wires up Voice Live params + (experimental) Azure semantic VAD config.
CONTEXT.md	Adds terminology glossary distinguishing Azure Speech vs Azure OpenAI Realtime vs Voice Live.
Cargo.toml	Adds the new service crate as a workspace member and workspace dependency.
.harper-dictionary.txt	Adds “VAD” to spelling dictionary.