Closes gsi-upm/senpy#34

Closes #22

* Improved debugging (back to using Flask's built-in mechanisms)
* Recursive model loading from json
* Added DEVPORT to Makefile
* Accept json-ld input. Closes #16
* Improved Exception handling in client
* Modified default plugin selection to only include analysis plugins
* More tests

* Added conversion plugins (API might change!)
* Added conversion to the analysis pipeline
* Changed behaviour of --default-plugins (it adds conversion plugins regardless)
* Added emotionModel [sic] and emotionConversion models

//TODO add conversion tests
//TODO add conversion to docs

* Added interactive debugging
* Better exception logging
* More tests for errors
* Added ONBUILD to dockerfile
  Now creating new images based on senpy's is as easy as:
  ```from senpy:<version>```. This will automatically mount the code to
  /senpy-plugins and install all dependencies
* Added /data as a VOLUME
* Added `--use-wheel` to pip install both on the image and in the
  auto-install function.
* Closes #9

Break compatibilitity:

* Removed ability to (de)activate plugins through the web

Bug-fixes and improvements:
* Closes #5
* Closes #1
* Adds Client (beta)
* Added several schemas
* Lighter string representation -> should avoid delays in the analysis
  with plugins that have 'heavy' attributes

Backwards-incompatible changes:
* Context in headers by default
* All schemas include a "@type" argument that is used for autodetection
  in the client

... And possibly many more, this is still <1.0

This should help with performance issues with models that have large
private variables.

* Added pre-commit: http://pre-commit.com
* Fixed flake8 errors
* Added flake8 pre-commit hooks
* Added pre-commit to Makefile
* Changed VERSION numbering
* Changed versioning to match PEP-0440

* Added debug Dockerfile/Makefile
* Validation of examples in docs

Also:
 * Shelve -> Pickle to avoid weird db problems
 * Serving schemas and contexts

Also:

* added drone-ci integration: tests for py2.7 and py3

There are also some slight changes to the JSON schemas and the use of
JSON-LD.

* setup.py:

Fixes #2

* The API has also changed, there are new parameters to send the
context as part of the headers.
* Improved tests
* PEP8 compliance (despite the line about gevent)

Moved from Yapsy again (it is not flexible enough), now we use a
custom solution.
The activation and deactivation of plugins is asynchronous, so
that plugins that take a long time don't interfere with the rest.