Scalability: Testing FDB scalability

To better understand how scalable FDB can be, we need a systematic way to evaluate and understand FDB's scalability.

This umbrella issue can be broken into the following sub-issues:
-[ ] Issue #2164: Identify the limits of each component and a cluster as a whole;
-[ ] Issue #2165: Emulate production workload at larger scales;
-[ ] Do performance test with production workload.

Start by reviewing sub-issues #2164 and #2165, then clarify the remaining production-workload performance-testing scope. Done means there is a systematic scalability evaluation covering component and whole-cluster limits, larger-scale production-workload emulation, and performance tests with production workload.

Written by the indexing model from the issue text.