Add Python 3.6, drop unsupported 2.6, 3.2, 3.3

[hugovk]

Reasons for dropping old ones

• https://en.wikipedia.org/wiki/CPython#Version_history

2.6

• https://snarky.ca/stop-using-python-2-6/
• http://www.curiousefficiency.org/posts/2015/04/stop-supporting-python26.html
• Current pip 9 deprecates Python 2.6 support, pip 10 won't support it (Drop support for Python 2.6? pypa/pip#3955)

3.2

• pip 8 no longer supports Python 3.2 (link)
• Coverage 4 no longer supports Python 3.2 (link)
• Requests no longer supports Python 3.2

3.3

• pip 10 will deprecate Python 3.3 support, pip 11 won't support it (Deprecate/Drop Support for Python 3.3? pypa/pip#3796)

[coveralls]

Coverage remained the same at 29.931% when pulling 988a240 on hugovk:update-versions into 6c465f7 on sixohsix:master.

[RouxRC]

For once, I'm not completely in favor of this: dropping 2.6 or 3.2 can potentially break existing deployed software on machines still relying on those (full disclosure, I run bots using python-twitter on old debian servers under python2.6 ;( )
Although I hear it is now officially deprecated, so I can understand endorsing it now makes sense, and bad server maintainers like myself should then probably just freeze their pip requirements on packages depending on this.
So my apologies for this time @hugovk, I'll respectfully let @sixohsix decide on this one ;)

[hugovk]

Here's how many pip installs Python Twitter Tools has had from PyPI over the past month:

$ pypinfo --pip --percent twitter pyversion
python_version percent download_count
-------------- ------- --------------
2.7              79.2%         17,371
3.6              10.0%          2,205
3.5               8.9%          1,945
3.4               1.6%            357
3.3               0.1%             29
3.7               0.1%             22
2.6               0.0%              9
3.2               0.0%              5

[sixohsix]

I'm with RouxRC on this. The arguments are not convincing to me.

Though users running 3.5 or 2.6 will probably experience pain trying to work with updated versions of pip, coverage, requests, etc. there is no reason to believe that they will encounter any problems with this twitter package. The code hasn't introduced any breaking changes so far as I know.

I would merge this change when either of two things happen:

• we introduce code changes that we know will break with 2.6 or 3.5.
• users encounter problems running this package with 2.6 or 3.5 (that is: when this happens, rather than put in the effort to fix this package, we might opt to deprecate those versions instead.)

In short, I do not want to merge this at this time.