Precompute and store visualization data object

[moreirathomas]

Once the report has been received from runner, we need to process it asap to produce vizualisation data (average, median, other stats,...). We need to parallelize this process. One possible solution is to send the work to a remote worker.

Possible infrastructure:

This implies:

• a nosql datastore to store the raw report waiting to be processed (redis, elasticsearch, memcached, some cgp alternative ?)
• a cloud worker capable of retrieving the raw data, processing it and storing the result in a clean database
• a sql database to store the processed vizualisation data

[moreirathomas]

Computing and storing of process stats is done by the worker. It lives in its own repo.
See benchttp/worker#2 and benchttp/worker#5.