Add stress test suite to concourse pipeline

[mattysweeps]

It is critical the Stackdriver-Nozzle works well with high volumes of logs. A stress test job should be added to the concourse pipeline.

• spins up X CF apps, each which log M/N bytes/message
• monitor and record VM health metrics for the Nozzle
• measure percentage of dropped logs
• measure log message delay
• measure Nozzle error count
• record all stats into a historical DB for later (trend) analysis