Support Perplexity API as a custom model (allow custom body fields)

Currently, it's impossible to use Perplexity's API as a custom model in TypingMind License version because Perplexity requires a title field in the request body — a non-standard OpenAI parameter that TypingMind cannot inject.

This limitation blocks a significant use case: combining TypingMind's excellent UI and workflow features with Perplexity's powerful web-search-augmented AI models (sonar-reasoning, sonar-pro, etc.).

Suggested solution:

Add an option to define arbitrary top-level body parameters in the Custom Model configuration (UI or JSON import), so users can include fields like "title": "typingmind" when needed.

Why this matters:

• Perplexity is one of the fastest-growing AI platforms, and many users want to consolidate all their AI models in TypingMind.

• Other tools (OpenCat, OpenRouter wrappers) already support this; TypingMind risks losing users to competitors that offer more flexible API configurations.

• This feature would also future-proof TypingMind for other APIs that may require non-standard body fields.

Current workaround status:

• TypingMind support confirmed there is NO workaround for License users today.

• The Perplexity Search plugin is helpful but limited to search results, not full chat completions.

I'd love to see this prioritized — it would make TypingMind the true "one UI to rule them all" for AI power users.

Thanks for considering!