Unify turn detection parameters

[pragmatrix]

No description provided.

[Copilot]

Pull request overview

Introduces a shared, provider-neutral context_switch_core::TurnDetection configuration to unify end-of-turn tuning across the Deepgram Flux and Microsoft Voice Live transcription services, with each backend forwarding only the fields it supports.

Changes:

• Added core::TurnDetection + ThresholdLevel and re-exported them from context_switch_core.
• Updated Deepgram and Voice Live transcribe params to accept the shared TurnDetection shape and mapped it to provider-specific SDK/session settings.
• Extended the examples/transcribe CLI and README with turn-detection tuning flags and usage examples; documented the design in an ADR and CONTEXT glossary.

Reviewed changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
services/microsoft-voice-live/src/transcribe.rs	Switch Voice Live params to use context_switch_core::TurnDetection and clarify forwarding behavior.
services/microsoft-voice-live/src/client.rs	Map neutral threshold_level/timeout_ms into Voice Live Azure multilingual semantic VAD config.
services/deepgram/src/transcribe.rs	Replace Deepgram-specific flattened turn-detection fields with shared TurnDetection (now nested).
examples/transcribe.rs	Add CLI flags for turn detection and thread a shared TurnDetection into provider params.
core/src/turn_detection.rs	New shared turn detection types (TurnDetection, ThresholdLevel).
core/src/lib.rs	Wire in and re-export the new turn detection module/types.
docs/adr/0001-turn-detection-provider-specific-fields.md	Document rationale/decision for provider-specific fields within a shared struct.
README.md	Add example commands for Deepgram and Voice Live end-of-turn tuning.
CONTEXT.md	Add terminology definitions for turn detection and threshold levels.