Prioritize external transcription endpoint over browser Web Speech API

Previously the on-device Web Speech API (Chrome/Edge/Safari) was always
preferred, and the configured external transcription endpoint was only
used as a fallback when the browser had no speech API at all (Firefox).

Flip the priority in Composer.tsx: if a transcription endpoint is
configured in Settings, use it for dictation regardless of browser
support, since it's generally far more accurate. Fall back to the
browser's on-device SpeechRecognition only when no endpoint is set.

Update the Settings hint text and README to describe the new behavior.
Rebuilt and redeployed to the repo root.

README.md

@@ -176,8 +176,11 @@ git push            # or however the server pulls the repo
 ## Notes & limitations
 
 - **Voice is client-side.** The API server has no `/transcribe` or `/tts`
-  endpoint, so dictation uses the browser's `SpeechRecognition` and replies use
-  `speechSynthesis`. Support varies by browser (best on Chrome/Edge/Safari).
+  endpoint. If a transcription endpoint is configured in Settings (any
+  OpenAI-compatible `/v1/audio/transcriptions` server), it's used for
+  dictation; otherwise dictation falls back to the browser's on-device
+  `SpeechRecognition`. Spoken replies always use `speechSynthesis`. Support
+  varies by browser (best on Chrome/Edge/Safari).
 - **Profiles.** The API server routes profiles via a `/p/<profile>/…` prefix, and
   each profile has its own `API_SERVER_KEY`. Add one connection per profile to
   see them all; the sidebar switches between connections.