Upgrade rails infrastructure

[eacharya]

• Move from single threaded MRI to multi-threaded Rubinius: Allows us to write code that leverages threaded designs
• Move from single-user Webrick to multi-user Puma: Allows us to use same server in both development and production version
• Refactor Rabbit connectors to start connection per worker by using on_worker_boot hook: Allows us to scale to multiple users/processes using rabbit

[eacharya]

Notes:

• Move to Rubinius postponed since don't have time to test if each ruby gem we use plays well with Rubinius. Theoretically, they should but most are tested only against default MRI so hesitant to jump right now. In any case, with multi-worker server, we should be already getting parallelization at the request level.
• Heatmap client inside rails moved to on_worker_boot hook inside puma config file. Other bunny servers extracted out to bin/data_ingest daemon. Wiki updated on running these background jobs.
• SSL support should be easy in puma - need to generate certificate and verify. Once puma-related issues have been identified (perhaps a week of usage), will introduce SSL. This will be done during cellroti server upgrade.

Merged to development: https://github.com/zigvu/kheer/commit/434fbcaf3524cd6e85edc30fbc40f663e14c1429