Composable output modes

Composable output modes

AIWG output modes provide a runtime-selectable, provider-neutral stack for voice, controlled-language, structure, and presentation constraints. The default is a true no-op named `unaltered`.

See `@$AIWG_ROOT/agentic/code/addons/voice-framework/docs/output-modes.md` for commands, precedence, composition and conflict semantics, protected literals, extension authoring, licensing, ASD-STE conformance rules, and migration from direct `voice-apply` usage.