it would be great if this could contain some information about the cost of these queries (number of processed documents, number of fields in query and so number of properties scanned).

This would be helpful to analyse what can be optimized in our usage.

This would also be useful if it could log failed queries / dataset requests.