feat(voice): render the voice control from a presenter

createVoiceControl now returns a presenter instead of building the
control with innerHTML: the session, cost, input, card and panel owners
write its state by part and listen for interactions, and VoiceControl.vue
renders it in the command dock, between the location bar and the visual
presets as before. Focus moves wait for the render; the waveform's bars
publish apart from the control and only when they change; the minutes
field keeps a draft until its owner writes it. Owner tests use a real
presenter on their own store, and view tests render the control.

CHANGELOG.md

@@ -20,6 +20,14 @@
   takes `omitViews` (for example `['provider-settings']`) for views an
   application leaves out (samehkhamis).
 
+- `./voice/control`'s `createVoiceControl({ panel })` returns the voice
+  control's presenter instead of element handles: owners write it by part
+  (`update('detail', { text })`) and listen with `on('button', fn)`, and the
+  command dock renders it from a store. `createVoiceCommands`' `createControl`
+  and `VoiceCardControls` (`{ ui }` instead of `{ root }`) take the same
+  presenter; the `reset` option is gone, since a newer control replaces the
+  previous one (samehkhamis).
+
 - `./scenes` presents the scene sharing dialogs and the EDIT DETAILS / SHARE
   SCENE bar through a store that the scene panels view renders, so a page
   that omits `layer-panels` shows neither (samehkhamis).