Get an understanding of infrastructure costs for taskbroker + workers

[markstory]

We need to have a more complete understanding of the infrastructure costs of taskbroker + taskworkers. Use the information collected in #127 and #126 to understand pod counts required for a given workload.

Given a workload of 100,000 tasks (that emulate production task runtime distribution):

• Measure the resources required to process a workload with celery
• Measure the resources required to process that same workload with taskbrokers.