Function compress_messages_with_ai 

pub async fn compress_messages_with_ai(
    messages: &[Message],
    config: &CompressionConfig,
    ai_mode: AiCompressionMode,
    fast_model: Option<Box<dyn Provider>>,
    token_usage: u32,
    context_window: u32,
) -> Result<Vec<Message>>

Compress messages with AI assistance (async version).

This is the new recommended API for compression with intelligent scoring, dependency tracking, and content summarization.