## Feature Request: WaveSpeed.ai Provider

**Is your feature request related to a problem? Please describe.** WaveSpeed.ai is a unified AI platform offering both **LLM inference** (93+ models via OpenAI-compatible API) and **AI media generation** (1,000+ models for text-to-image, image-to-video, text-to-video, audio). Currently, LiteLLM does not support WaveSpeed.ai as a provider, requiring users to configure it manually as a custom endpoint.

**Describe the solution you'd like** Add WaveSpeed.ai as a first-class provider in LiteLLM for both:

### LLM Service - OpenAI-compatible chat completions API at `https://api.wavespeed.ai/v1/chat/completions` - Model names follow the pattern: `{provider}/{model}` (e.g. `openai/gpt-5.6-luna`, `anthropic/claude-sonnet-4.6`, `deepseek/deepseek-v4`, `google/gemini-2.5-flash`, etc.) - Supports vision, tool use, and prompt caching - API key authentication

### Media Generation (Image/Video/Audio) - REST API for predictions at `https://api.wavespeed.ai/v1/predictions` - Models include: Seedream 5.0, Seedance 2.0, FLUX 2, WAN 2.x, Kling, Luma, GPT-Image, etc. - Asynchronous prediction workflow (submit → poll/get result) - Supports webhooks for comple…