Create additional benchmarking for first public rollout

[szabolcsmaj]

The current benchmarks are not that satisfactory. A simulation with a more complex project is needed to assure that the overhead GhostWriter adds is negligible.

We should create a stress test by bombarding the API with calls. The key is to test the system with and without the instrumentation while it's working normally, meaning no exception happens.