Add support for multi-cluster runners

[b-deam]

Use cases involving multiple clusters like Cross Cluster Search (CCS) or Cross Cluster Replication (CCR) require various additional runners in order for tracks to properly set them up,

We should implement dedicated runners in Rally for;

1. Configuring 'local' and 'remote' cluster (these terms are relative) seeds
2. Add support for 'local' and 'remote' clusters to 'follow' indices

Additionally, all current runners are written only to execute in the context of a single Elasticsearch client for the first cluster passed in via target-hosts (i.e. default). This present a problem for when users would like operations to execute across multiple clusters. We should consider adding a new operation (e.g. all-clusters) that will run a given operation on all clusters without requiring us to modify each existing runner.

Relates;
#1563
elastic/rally-tracks#293

[b-deam]

Upon review, multi-cluster benchmarks are almost always going to involve complex enough setups and architectures that it will be difficult to abstract cleanly via generalised multi-cluster runner(s). Our docs suggest as much too.

Closing this for now.