✨ Bulk fetch indicators for inference runs

Bulk fetch default indicators for multiple inference runs at once.

Returns aggregated metrics for up to 20 completed runs in a single request, used by the Comparison Matrix to show all versions side by side.