Rename or change metrics meaning for 2 groups of metrics: client.*, cluster_grinder.*

[ikopylov]

cluster_grinder is better to rename to cluster.
The main problem is with the client.* metrics. It is not clear what these metrics are for. I think, it is better to change the name and meaning of these metrics. To track performance, we are interested in getting metrics for the entire range of requests for a node (regardless of the force_node flag). So it would be good to track the values for the whole incoming put/get/exist/delete operations, not just for the case when force_node == true. And the metrics in this case can be names as node. Basically, that means that node = client + cluster_grinder