Split out the redis workload graphs into primary vs replica

Because of some recent additional work (bigkeys data extract) that runs on the replica, the odd replica behavior (that is not necessarily a problem) can obscure the data from the primary, which is much more important. Splitting out those graphs makes them easier to interpret

Snapshot available for a limited time at https://dashboards.gitlab.net/dashboard/snapshot/Bg6U6IBGrjvhCHyIFFA9BqCEH3b2pQK6?orgId=1

Merge request reports

Loading