[performance testing] Formalize hardware conditions for performance testing

[spalger]

Performance testing requires a stable environment to reduce the number of variables affecting the final results. We need to formalize the requirement to "baseline" hardware setup for performance testing, reflecting the configuration relevant for most users (or worst-case scenarios)

CPU

Nodejs application runs on a single thread, v8 spawns other threads for internal needs: libuv, garbage collection, etc.
Probably, we should test for a worst-case scenario when the Kibana application is run on an instance with a single CPU?

RAM

Available RAM defines how much memory can be allocated to the nodejs process (V8's old memory section). Which, in turn, determines how often the runtime engine initiates the garbage collection.
We can get the most frequently used available RAM value from our telemetry.

Network bandwidth

Network conditions are crucial for perceived performance. We need to formalize what network conditions are the baseline for performance testing. We can get this information from the Elastic Cloud setup.

[elasticmachine]

Pinging @elastic/kibana-operations (Team:Operations)

[spalger]

We're using static bare-metal machines now.