Rewrite the library to use the Watson Developer Cloud Python SDK

[ResidentMario]

This library currently relies on the custom event_insight_lib.py for calling to the IBM Watson API. It would be more secure and more stable (and possibly also faster) to use instead the Concept Insights module in the IBM-underwritten Watson Developer Cloud Python SDK (which is not yet finished).

[ResidentMario]

This would also (hopefully) address #6.

[ResidentMario]

Once the SDK is released I can tackle an overhaul of the underlying framework that should address two outstanding performance issues:

• Multithreading --- right not everything runs on a single thread, if I split up the API calls and organized them into multiple threads I could get an order of magnitude faster performance. Really looking forward to seeing the before and after!
• Batch requests --- edge generation methods fail for ConceptModel objects with more than 50 (hard limit? soft limit? not sure) concepts because the API refuses get_relation_scores() calls with too many parameters. I should rewrite those methods to send those requests in batches of 50 (or whatever the limit is).
• Key assignment --- right now the service has to re-validate the key every run-time (and occasionally generate a new one on an hourly basis), this will no longer be necessary once I start using the SDK's direct connection validation.
• Better encoding --- see #11.

[ResidentMario]

Re: Batch requests

After some more experimentation it looks like this is a soft limit on the length of the URL string that is passed to Watson?