Subprocess32

-----------------

* Fixed the packaging problem where the stdlib python3_redirect shim is
supposed to be installed on Python 3.
* Renamed _posixsubprocess to _posixsubprocess32 for consistency.
* Unset CLOEXEC on file descriptors given to Popen pass_fds. (GH 4)
* Drop support for Python 2.4 and 2.5.
* Adds a configure script - run by setup.py - to supply the missing feature
define's for the _posixsubprocess32 extension module for the things that
Python 2's own configure generated pyconfig.h does not already provide.

-----------------

-----------------

* Backport the new subprocess.run() API from Python 3.5.
* Backport subprocess.DEVNULL support from 3.3.
* Allow stderr to be redirected to stdout even when stdout is not redirected.
https://bugs.python.org/issue22274
* Fix subprocess.Popen.wait() when the child process has exited to a
a stopped instead of terminated state (ex: when under ptrace).
https://bugs.python.org/issue29335
* Include the private API needed by the multiprocessing module for people who
want to drop subprocess32 in as a replacement for their standard library
subprocess module.
* Fix a compilation issue regarding O_CLOEXEC not being defined on ancient
Linux distros such as RHEL 5.

-----------------

-----------------

* Issue 6973: When we know a subprocess.Popen process has died, do
not allow the send_signal(), terminate(), or kill() methods to do
anything as they could potentially signal a different process.
* Issue 23564: Fixed a partially broken sanity check in the _posixsubprocess
internals regarding how fds_to_pass were passed to the child. The bug had
no actual impact as subprocess32.py already avoided it.

-----------------

-----------------

* Moved the repository from code.google.com to github.
* Added a _WAIT_TIMEOUT to satisfy the unsupported people entirely on
their own trying to use this on Windows.
* Similarly: Updated setup.py to not build the extension on non-posix.

-----------------

-----------------

* Issue 21618: The subprocess module could fail to close open fds that were
inherited by the calling process and already higher than POSIX resource
limits would otherwise allow. On systems with a functioning /proc/self/fd
or /dev/fd interface the max is now ignored and all fds are closed.

-----------------

-----------------

* Fixes issue 21291: Popen.wait() is now thread safe so that multiple
threads may be calling wait() or poll() on a Popen instance at the same time
without losing the Popen.returncode value.
* Fixes issue 14396: Handle the odd rare case of waitpid returning 0 when not
expected in Popen.wait().
* Fixes issue 16962: Use getdents64 instead of the obsolete getdents syscall
on Linux. Some architectures do not implement the latter.

-----------------