vision: OpenAI provider array content format for images

[bug-ops]

Parent: #490
Depends on: #491

Scope

Implement vision support in the OpenAI provider.

Changes

1. crates/zeph-llm/src/openai.rs
   • Override supports_vision() -> true
   • When message contains MessagePart::Image: serialize content as array [{type: "text", text: "..."}, {type: "image_url", image_url: {url: "data:mime;base64,..."}}] instead of plain string

Notes

• GPT-4o and later models support vision natively
• Images sent as base64 data URLs in the content array
• Compatible provider (crates/zeph-llm/src/compatible.rs) may also benefit — check if it shares the OpenAI request format