Scaling MAGI

[isi-lincoln]

So this is to talk / track the concept of making MAGI a bit more modern.

The idea being is that each module is a service, services expose RESTful API. MAGI takes with the module to implement the functionality. Modules (services) are self-contained, handle error checking, and wont cause MAGI to crash.

Some amount of code rewrite is necessary. The idea is if we can gut MAGI to keep only the necessary state, limit the amount of overhead, and access modules via API. In the process make it so MAGI can be testbed independent (which as I understand it - it has hardcoded values for DETER)

[isi-lincoln]

Per discussion with Geoff we may want to go this way, but also look at non-MAGI agents that are services such as https://github.com/ISIEdgeLab/distributed_agents, so ideally - changing MAGI to communicate with the distributed agents rather than "modules"