Tagging version 1.8.4 release of SPARQLWrapper.

Tagging version 1.8.3 release of SPARQLWrapper.
Problably it is not working for Python3

• Fixed bug (#100)
• Updated doc
• Added Unauthorized exception in SPARQLWrapperExceptions
• Added support for custom HTTP headers (#52)
• Changed timeout setting (#106)

• Update classifiers (Python 3.6)
• Added some documentation about the parameter to indicate the output format
• Fixed typo in width calculation
• Added support for CSV, TSV (PR #98)
• Added support for Only HTTP Content Negotiation (#82)

• Updated return formats for not content negotiation situations
• Included license in the MANIFEST (issue #76)
• Added explicit support for RDF/XML as allowed format (issue #75)
• Added proper shebang (issue #78)
• Moved keepalive as optional dependency (issue #79)
• Fixed hash check on prefixes (issue #77)
• Fixed epydoc warnings (issue #41)

• Removed wrong response encoding (issue #70)
• Authorization header bug when using Python 3 (issue #71)

• Removed pip dependency on setup (issue #69)

• Fixed packaging (issue #66)

• Finally fixed the keepalive issue in all Pyhon versions (issue #65)
• Removed old JSON layer in favor of the native json module

• Moved to the new keepalive package (issues #53 and #61)

• Fixed build in Python 3.x (issue #57)

• Added support to HTTP Digest Auth Support (issue #45)
• Improved print_results showing language tag (xml:lang) and datatype
• Updated to RDFLib 4.x

• Fixed unicode problems on setup (issue #42)

• Fixed unicode problems with urllib in Python 3 (issue #35)
• Restored SPARQLWrapper2 class (issue #36)
• Enhanced warning for missing rdflib-jsonld (issue #38)
• Fixed build system (issue #39)

• Fixed query type detection with comments (issue #32)

• Added missing query types (issue #17)
• Added a new method to the API to select the request method to be fully SPARQL 1.1 Protocol compliant (issue #28)
• Improved the test suite coverage, including support to run the tests under Python 3.x (issues #20, #24 and #31)

• Returning raw response in case of unknown content type returned
• Fixed some issues with the last version of the SPARQL 1.1 Update Protocol
• setQuery() doesn't imply resetQuery() anymore
• Deprecated addCustomParameter(), addParameter() and clearParameter() come to provide all required functionality
• SPARQLWrapper, QueryResult, Value, Bindings (and classes inherited from them) are new-style classes now
• POST queries are accompanied by full set of parameters now
• Added rudimentary support for JSON-LD
• Added proper unit tests without dependencies of external endpoints
• Fixed Python 3 compatibility issues in SmartWrapper module

• Implemented update operation according the latest SPARQL 1.1 Protocol drafts (i.e., switching to 'update' parameter)

• Added the possibility to use two different endpoints for reading and writing operations
• New print_results() function for users testing

• Update handling 500's coming from SPARQL endpoint (feature request #3198363)
• Added Python 3.x support (feature request 3022722)
• Warning when returned format would be different than the requested one

• Updated for working with RDFLib3 too (feature request #3117442)
• fixed bug with prefixes' regex (#2320024)

• Supporting keep-alive in SPARQLWrapper if urlgrabber is available (ticket #2929881)
• fixed bugs (#2949834)

• Added some support for SPARUL
• Improved HTTP related code
• Many other minor bugs fixed

• Remove pyxml dependency. Instead, use xml.dom.minidom
• Updated setup installation (added rdflib dependency)
• Updated example.py (added XML, N3 and RDF examples)

• Remove simplejson dependency for python => 2.6 version
• Added feature to choose the json module to use

• Added a new method to add custom parameters (deprecated old way to do it)

• Updated setup installation
• Patched to work with JSON in Python>=2.6

• Allowed non-standard extensions (such as SPARUL).
• Exceptions fixed.
• Added another example.