Components.AI: Implement a reusable microphone composer part

### Is your feature request related to a problem? Please describe the problem.

Applications must implement voice acquisition and composer integration themselves.

### Describe the solution you'd like

Provide a standalone microphone input part with recording/listening state, permission and unsupported handling, cancellation, and transcription appended to current composer text without destroying an existing draft.

### Additional context

Leaf task for #68227 requirement C4. Depends on the composer context and minimal JS module.

Start with the composer context and minimal JS module identified in the issue, then review requirement C4 in the linked parent issue #68227. The work is complete when a standalone microphone input handles recording/listening state, permissions, unsupported cases, cancellation, and appends transcription without destroying existing composer text.

Written by the indexing model from the issue text.