Include size values for cluster state

[F21]

When getting the status using _status, the size of each index is given. However, the translog and cluster state's sizes aren't given. This means that we cannot really tell how much space is being occupied by the index on disk.

It would be nice if these 2 pieces of information can be included in the response for each index in _status, so that we can calculate the exact disk space the index occupies.

[rpedela]

+1

I would actually like to know how much disk space is used in total including the stored data.

[clintongormley]

The translog size is now available in the nodes-stats API.

[jpountz]

Upgrading to adoptme: let's provide the size of the compressed cluster state in our monitoring APIs. NOTE: this would be the size of the compressed serialized representation of the cluster state that is typically used for inter-node communication, not the amount of memory that the cluster state occupies on each node.