docs: Correct continuous mode disclaimer to match actual behavior

[untemps]

Summary

The README Disclaimer claimed continuous mode is for "real-time transcription and always-on voice command use cases." Both claims contradict the actual behavior and the README's own API table. This reword describes what continuous mode really does and drops the false framing.

Changes

• README.md (Disclaimer): replace the "real-time transcription and always-on voice command use cases" sentence with an accurate description — continuous mode keeps the recognition session open across speech segments and delivers a single, aggregated transcript through onResult when the session ends (stopped by a second click or by silenceTimeout), and commands are not evaluated in continuous mode. Keeps the pointer to the Vocal component API table.

Why the old wording was wrong

• No real-time transcription: useVocal builds the instance without interimResults (src/hooks/useVocal.ts:43); @untemps/vocal emits a single aggregated event just before end, so onResult fires once with the full session transcript — not incrementally.
• No always-on voice commands: command matching is gated off in continuous mode (src/components/Vocal.tsx:251, if (!continuousRef.current)). The API table (README.md:278) already states commands are not evaluated in continuous mode.

The new wording reuses the API table's exact verbs so the Disclaimer and the table no longer contradict each other.

Test plan

• yarn typecheck — passes
• yarn lint — passes
• yarn build — passes
• yarn test:ci — 189/189 passing
• Read the rendered Disclaimer and confirm it no longer mentions real-time transcription or always-on commands, and matches the Vocal component API table.

Closes #248