Sync ZenMux models with latest website data

[LYY]

This PR synchronizes the ZenMux provider models with the latest data from zenmux.ai.

Changes Summary

New Models Added (17)

Provider	Model ID	Description
google	gemini-3-pro-image-preview	Gemini 3 Pro with image capabilities
google	gemini-3.1-flash-lite-preview	Cost-efficient Gemini 3.1 model
minimax	minimax-m2.7	High-performance software engineering model
minimax	minimax-m2.7-highspeed	Faster variant of M2.7
openai	gpt-5.3-chat	GPT-5.3 for chat use cases
openai	gpt-5.3-codex	GPT-5.3 optimized for coding
openai	gpt-5.4	OpenAI's frontier model
openai	gpt-5.4-mini	Efficient variant of GPT-5.4
openai	gpt-5.4-nano	Ultra-efficient GPT-5.4
openai	gpt-5.4-pro	Advanced GPT-5.4 variant
qwen	qwen3.5-flash	Fast inference model
qwen	qwen3.5-plus	Enhanced Qwen 3.5
volcengine	doubao-seed-2.0-code	Code-optimized Doubao model
x-ai	grok-4.2-fast	Fast Grok 4.2 variant
x-ai	grok-4.2-fast-non-reasoning	Non-reasoning Grok 4.2
xiaomi	mimo-v2-omni	Multimodal interaction model
xiaomi	mimo-v2-pro	Professional MiMo variant
z-ai	glm-5-turbo	Fast GLM-5 inference

Deprecated Models (1)

Model	Reason
anthropic/claude-3.5-sonnet	No longer listed on ZenMux website

Data Source

• Website: https://zenmux.ai/models
• Filter: sort=newest, supported_protocol=messages
• Parsed using: lightpanda browser

Validation

• All TOML files validated against schema
• Pricing data extracted from website
• Context limits and capabilities accurately reflected

Notes

• Total ZenMux models after this PR: 87
• Models not found on website have been marked as "deprecated" with status field