The contents of this attribute are an implementation detail, as
documented for #9442, so we should not parse it, to support non-CPython
VMs in the future.

Unfortunately, one use comes directly from PEP 345, so an edit will have
to be agreed before fixing the code (see comment in d2.markers).

Other remaining uses are found in d2.compiler and could be replaced by
the platform module (which also parses sys.version, but then it wouldn?t
be my fault :)

pysetup create used to convert package_data from an existing setup.py
into extra_files, but these files are only present in sdists, not
installed: they don?t have the same use case at all, so converting one
into the other did not work.

This moves individual backported functions from compat into a new
_backport.misc module.  compat regains its original purpose of
containing 2to3 support code (i.e. compat for projects using distutils2,
not internal compat for d2 code) and all backports now live under the
private _backport subnamespace.

Also fix the definition of ?any? to bind it to the builtin in 2.5+
(fsencode and detect_encoding not changed, they never exist in
2.x).

This should help avoiding issues like #13170.

Because of implicit relative imports, _backport.tarfile already imported
shutil from the backports instead of the stdlib; I have inlined the only
function it needed to remove the circular dependency between these
modules.

In addition, update the create module to use logging.

- Remove __main__ blocks obsoleted by pysetup
- Fix typo ?seperate?
- Add one test that was promised but not written
- Reorganize one file

shutil, sysconfig, tarfile and their tests have been updated to the
latest 3.2 version (except for test_tarfile which is not backported yet)
and edited to be compatible with 2.4.

Duplicates added in util during the Great Update have been deleted, as
well as functions I removed recently in packaging.  Unneeded modules in
_backport have been deleted or moved to d2.compat.

Backports:
- sysconfig is now always imported from our backports
- when hashlib is not found, our backport is used instead of the md5
  module (debatable; we could just drop hashlib)

Version-dependent features:
- PEP 370 features are only enabled for 2.6+
- the check for sys.dont_write_bytecode was fixed to use getattr
  with a default value instead of hasattr

Idioms/syntax:
- octal literals lost their extra 0
- misused try/except blocks have been changed back to try/finally
  (it?s legal in 2.4 too, it?s only try/except/finally that isn?t)
- exception catching uses the regular 2.x idiom instead of sys.exc_info
- file objects are closed within finally blocks (this causes much
  whitespace changes but actually makes diff with packaging easier)

Renamed modules:
- some missed renamings (_thread, Queue, isAlive, urllib.urlsplit, etc.)
  were fixed

Other:
- a few false positive replacements of ?packaging? by ?distutils2? in
  comments or docstrings were reverted
- util.is_packaging regained its name
- assorted whitespace/comment/import changes to match packaging

The print statement fixes that were in my branch override the print(u'')
calls added by Alexis; the import at function scope were removed for the
usual reasons; d2._backport.hashlib is used if hashlib is not available
instead of md5.