Combine Knowledge API with Evaluation Engine

[mdfalkow]

Currently the APIs for knowledge querying and evaluation are separate (adapt-api and adapt-reasoner), as shown in the architecture diagram below:

The current layout has two separate services requesting from the same data layer, which is an antipattern that was unintended when I originally was designing the system (I had not considered that the reasoner would interface with TWKS directly).

ADAPT could be streamlined by combining Knowledge API and Evaluation Engine into a single Java-based API. This approach would remove the potential bottlenecks, and reduce container size.

[mdfalkow]

If TWKS implements a browser-based Javascript client (see tetherless-world/twks#73) then there is possibility for aggregation of system logic in the front-end (which is less desirable for scaling reasons, but still an option nonetheless).

In that scenario evaluation engine logic would still remain in a separate service, as there is no JavaScript-based implementation of the HermiT reasoner (as far as I know). Hence we would have 3 services instead of 4; however, the resulting system would be the same diamond-shaped antipattern as the current system.

This gives more reason to moving to a Java-based API.