feat: add Novita AI as optional LLM provider

[Alex-wuhu]

What this PR does

Adds Novita AI as an optional, drop-in replacement for existing OpenAI-compatible providers.

Usage: Set NOVITA_API_KEY in your environment — the existing configuration is preserved and unchanged.

Novita provides an OpenAI-compatible API endpoint (https://api.novita.ai/openai) with cost-effective access to leading models including DeepSeek, Llama, and more.

Changes

• Created src/lib/models/providers/novita/ directory.
• Implemented NovitaProvider, NovitaLLM, and NovitaEmbedding using the OpenAI-compatible pattern.
• Registered NovitaProvider in src/lib/models/providers/index.ts.
• Updated README.md to include Novita AI in the list of supported providers.

Integration analysis

• Architecture: multi_provider
• Pattern reference: src/lib/models/providers/groq/
• Deviations: Added embedding support via NovitaEmbedding extending OpenAIEmbedding.

Testing

Set NOVITA_API_KEY=<your-key> and you should be able to select Novita AI as a provider in the setup screen and use its models for both chat and embeddings.

---

Summary by cubic

Adds Novita AI as an optional, OpenAI-compatible provider for chat and embeddings. Enables using Novita models by setting NOVITA_API_KEY, with no other config changes.

• New Features

  • Added NovitaProvider with chat and embedding support via https://api.novita.ai/openai.
  • Registered provider in the models index and updated README.
  • Fetches model list from /openai/models for selection in the UI.

• Migration

  • Set NOVITA_API_KEY in the server environment.
  • Select Novita in the setup screen and choose models for chat and embeddings.

^{Written for commit 826e363. Summary will update on new commits.}