Parallelization of execution support

[andrewvc]

We should consider strategies for running larger test suites. For truly large workloads #277 will help scale things out, but before we get to that we should consider maximizing resources on a single machine.

Ideally we would modify the runner to:

1. Detect the number of CPU cores, running N-1 tests simultaneously
2. Figure out a strategy to prevent resource intensive tests from impacting perf results from its test neighbors (CPU pinning?)

[paulb-elastic]

Linking to related issue elastic/beats#23216

[andrewvc]

After talking to paul, I think this probably makes more sense to do in beats rather than synthetics.