Add the ability to track internally generated and externally generated load separately

There are a variety of sources of internal load on a cluster, and this can often cloud discussions with cluster users about their workloads. It also makes it harder to determine the effects of various internal processes.

It would be useful if we could have some way to distinguish the load generated by internal activities (e.g. data distribution, backup/DR, consistency check, etc) from those generated by clients of the database. Perhaps even better from a tooling standpoint would be a more general way to distinguish internal transactions by category (so we could, for example, understand the effects of DD by itself).

This may require us to mark each request with additional information about the source of the traffic so that the process handling the request can increment the appropriate counters. In that case, we'll have to balance the functionality with the cost of sending the required information.

No files, tests, or entry points are named. Start by locating request handling and accounting for data distribution, backup/DR, consistency checks, and client traffic; done would mean these sources can be distinguished, ideally by internal category, while keeping the added request information cost acceptable.

Written by the indexing model from the issue text.