feat(llm): add NEAR AI Cloud provider

[PierreLeGuen]

What does this PR do?

Adds NEAR AI Cloud as an LLM provider across OpenVoiceUI's provider surfaces:

• Registers a native nearai LLM provider using NEAR AI Cloud's OpenAI-compatible chat completions API.
• Exposes NEARAI_API_KEY in setup, generated OpenClaw config, admin provider metadata, and profile/provider UI surfaces.
• Documents NEAR AI Cloud TEE inference configuration for users.

Why?

This gives OpenVoiceUI users another OpenAI-compatible LLM provider option, with NEAR AI Cloud's TEE-backed private inference available through the existing gateway/provider setup.

Type of change

• Bug fix
• New feature
• TTS provider (new or improved)
• Frontend / UI
• Documentation
• Refactor / cleanup

Testing

• venv/bin/python -m pytest tests/ --tb=short -q
• node --check setup-config.js
• node --check install.js
• Setup generator smoke test with NEARAI_API_KEY confirmed generated OpenClaw provider config, auth profile, and .env entry.

Checklist

• Tested end-to-end with live voice input
• No API keys or secrets in my changes
• .env.example updated for NEARAI_API_KEY
• No new Python packages required

[MCERQUA]

Thanks for this, @PierreLeGuen — it's a clean, well-tested PR and the NEAR AI Cloud confidential-inference angle is genuinely interesting.

We've decided not to merge individual provider integrations one-by-one, though. OpenVoiceUI is moving toward a generic 'add any OpenAI-compatible provider' system (see #244 / #246) — a single place where users configure a provider's base URL, key, and models without code changes. Once that lands, NEAR AI Cloud will be addable by any user who wants it (with their own key), no per-provider PR needed — and it keeps us from hardcoding each vendor across the codebase. In the meantime, NEAR works today via custom OpenClaw provider config locally.

Closing in favor of that direction — but genuinely appreciate the contribution and the quality of it. 🙏