lotus(#3): soundboard audio injection via widget action

Adds io.lotus.inject_audio (toWidget): mixes a soundboard clip into the
call so other participants hear it. Publishes the clip as a separate
Unknown-source LiveKit track (rendered by MatrixAudioRenderer) rather than
splicing into the mic track, so the denoise pipeline is untouched; the
track is unpublished when the clip ends (with a 30s safety cap). This is
the real call-audio injection that was impossible against the prebuilt EC
bundle. Additive: no-op unless the host sends the action.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

src/room/InCallView.tsx

@@ -31,6 +31,7 @@ import { useCallViewKeyboardShortcuts } from "../useCallViewKeyboardShortcuts";
 import { widget } from "../widget";
 import { startLotusCallState } from "../lotus/lotusCallState";
 import { startLotusFocus } from "../lotus/lotusFocus";
+import { startLotusAudioInject } from "../lotus/lotusAudioInject";
 import styles from "./InCallView.module.css";
 import { GridTile } from "../tile/GridTile";
 import { SettingsModal, defaultSettingsTab } from "../settings/SettingsModal";
@@ -287,6 +288,9 @@ export const InCallView: FC<InCallViewProps> = ({
   // [lotus] Handle the host's io.lotus.focus_participant action to pin a
   // participant to the spotlight (#4). No-op unless the host sends it.
   useEffect(() => startLotusFocus(vm), [vm]);
+  // [lotus] Handle the host's io.lotus.inject_audio action to mix a soundboard
+  // clip into the call as a separate track (#3). No-op unless the host sends it.
+  useEffect(() => startLotusAudioInject(vm), [vm]);
 
   const fatalCallError = useBehavior(vm.fatalError$);
   // Stop the rendering and throw for the error boundary