[profiler] Research on autoscaling for the rdb process

[zanbuchanan]

This will be part of the new profiler process.

Building of the actual proccess will be done as part of #42

When the rdb is expected to run out of memory (part of an algorithm that will be developed as part of this ticket) it will start up another rdb with greater memory, wait for it to start and register itself. The original rdb will then deregister itself, write, and kill itself.

Goals:

1. Create an algorithm for when the rdb will need to increase or decrease it memory. MVP will be to look at the past 5 days of memory usage.
2. Have the rdb spin up another rdb and kill itself