Develop a stress test tooling

[ricardomaraschini]

We need a stress test tooling. The idea is to be able to run a full stress test from a single machine, just pointing to a cluster where tagger is already running.

Information to be collected:

1. Memory usage average
2. Number of open file descriptor (check if we are not leaking when under pressure)
3. CPU usage average

Out of this task come out with good values for adding into tagger Deployment "requests" and "limits".