Add ElevenLabs Scribe as an optional Voice Engine for dictation

[maxurby]

I’d like to use ElevenLabs Scribe for speech-to-text while keeping FluidVoice’s existing dictation workflow. Could FluidVoice offer ElevenLabs Scribe as an opt-in choice alongside the local models in the Voice Engine picker?

Proposed behavior

• Select ElevenLabs Scribe and enter your own ElevenLabs API key.
• When Scribe is selected, FluidVoice sends the recorded dictation audio to ElevenLabs for transcription and inserts the resulting text as usual.
• Label Scribe as a cloud engine so users know their audio will be sent to ElevenLabs. Keep local engines available as the default option.

For a first version, transcribing after the recording stops would cover this use case.

Discussion #615 proposes an OpenAI-compatible remote endpoint. This request is for direct integration with the ElevenLabs speech-to-text API.