multilingual voice input

### Describe the feature or problem you'd like to solve

Copilot CLI voice input only supports built-in English and Spanish speech-to-text models. I’d like users to be able to choose or configure their own STT model and language.

### Proposed solution

I’d like to suggest an improvement to the voice input feature in GitHub Copilot CLI. Right now, users are limited to the speech-to-text models that are built into the product, which makes the feature less useful for people who need support for other languages, such as Portuguese.

My suggestion is to allow users to configure or specify their own transcription model, instead of being restricted to the default set of available models. For example, this could include:
- selecting a local model by path, identifier, or configuration;
- supporting additional languages such as pt-BR and pt-PT;
- choosing a default model per language;
- preserving local processing for privacy.

This would make voice input much more flexible and inclusive for multilingual users and for language communities that are not currently well supported. In my case, as a native Portuguese speaker, dictating in English or Spanish reduces transcription quality and makes the experience less natural.

### Example prompts or workflows

_No response_

### Additional context

_No response_

No files, tests, or entry points are named. Start by locating the CLI voice-input and speech-to-text implementation, then determine how model and language selection are currently handled. Done should cover configurable models, Portuguese language support, per-language defaults, and preservation of local processing.

由索引模型根據 Issue 內容生成。