[Refactor]: worker as a separate entry point for cluster primary

[snewcomer]

ref #833

This PR is intended to be an initial migration towards a vision for the fastboot app server that ensures extensibility and reliability.

What this PR includes

• Two separate entry points: cluster setup with separate worker entry point - docs. This reduces conflation between primary and worker APIs as well as ensure all application state and information is the responsibility of the worker. Moreover, it moves us to completely exposing the worker to end user configuration and setup.
• HTTP server should be cluster worker concerns and not passed as top level to primary process.

What this PR does not include (future follow on PRs)

• Exposing worker for configuration and setup script
• extend message handling to more than a hard coded set of messages
• Event emitter API to receive messages from http server or primary; bi directional communication
• Further improvements to download and notify API
• Middleware should be loaded via DAG.

See repo for ideas included in this PR.

https://github.com/snewcomer/ember-alt-fastboot-app-server