Add context manager to Reader and other improvements

[megies]

Hi there, this PR proposes to add a context manager that makes sure open files are closed properly in the most Pythonic way there is (also see #107).

I've also made some other improvements (at least I think they are ;-), these can be singled out if you don't agree with all of them:

• derive all classes from object (i.e. make them new style classes)
• add convenience mapping "shapeType number" <---> "shapeType string" (keeping old module attributes like shapefile.POINT)
• improve closing of individual pieces dbf/shp/shx, I think it's not safe right now, if one of these fails, the rest of them are not closed by Reader.close()

I've just started using pyshp for shapefile manipulations, thanks for the project! It's so convenient that it's noarch, pure Python and py2/3! ❤️

Eventually tests should really be implemented as unit tests, which is not a lot of work when relying on pytest, and Travis (ah Travis is set up already) and Appveyor should be set up as CI. I don't have time for this right now but might make a PR for this in the future..

Fixes #121

[karimbahgat]

Very useful, should make file handling more stable. Thanks 👍