Both Python 2 & Python 3 support representing Unicode strings with a u
prefix. Can drop the _u() compat shim and let the interpreter handle it.

If x.isspace() is True, then unidecode(x).isspace() should be True as well
(unless it is an empty string)

Python 2.6 and 3.3 are end of life. They are no longer receiving bug
fixes, including for security issues. Python 2.6 went EOL on 2013-10-29
and 3.3 on 2017-09-29. For additional details on supported Python
versions, see:

https://devguide.python.org/#status-of-python-branches

Removing support for EOL Pythons will reduce testing and maintenance
resources.

Using pypinfo, here are the download statistics for Unidecode over the
last 30 days, showing minimal 3.3 & 2.6 installs.

$ pypinfo --percent unidecode pyversion

| python_version | percent | download_count |
| -------------- | ------- | -------------- |
| 2.7            |  66.07% |        244,263 |
| 3.6            |  19.69% |         72,777 |
| 3.5            |   8.27% |         30,585 |
| 3.4            |   5.59% |         20,663 |
| 3.7            |   0.29% |          1,062 |
| 2.6            |   0.06% |            210 |
| 3.8            |   0.02% |             72 |
| 3.3            |   0.01% |             44 |
| 3.2            |   0.00% |              6 |
| None           |   0.00% |              2 |

Also, add unidecode_expect_nonascii. "unidecode" is now an alias for
"unidecode_expect_ascii"