Function request_incremental_grouping 

pub async fn request_incremental_grouping(
    backend: LlmBackend,
    model: &str,
    summaries: &str,
) -> Result<Vec<SemanticGroup>>

Request incremental grouping: assign new/modified hunks to existing groups or create new ones.

The summaries parameter already contains the existing group context prepended (from incremental_hunk_summaries), so we just need a different system prompt.