* [Fixed] Keep setting option for the voice
- It was always changed to the first available voice
- Also added an error if the selected voice isn't valid
* [Fixed] elevenlabs_tts API key handling
- The one from the settings wasn't applied
- We always got "Enter your API key", even when the settings specified
an api_key
* [Added] elevenlabs_tts model selection
- Now we can also use the "eleven_multilingual_v1" model.
Used for anything but english.