Without this function people would be tempted to use the other date functions
in email.utils to compute an aware localtime, and those functions are not as
good for that purpose as this code.  The code is Alexander Belopolsy's from
his proposed patch for issue 9527, with a fix (and additional tests) by Brian
K. Jones.

This is a danger of focusing on unit tests: sometimes you forget
to do the integration tests.

When the new policies are used (and only when the new policies are explicitly
used) headers turn into objects that have attributes based on their parsed
values, and can be set using objects that encapsulate the values, as well as
set directly from unicode strings.  The folding algorithm then takes care of
encoding unicode where needed, and folding according to the highest level
syntactic objects.

With this patch only date and time headers are parsed as anything other than
unstructured, but that is all the helper methods in the existing API handle.
I do plan to add more parsers, and complete the set specified in the RFC
before the package becomes stable.

This patch primarily does two things: (1) it adds some internal-interface
methods to Policy that allow for Policy to control the parsing and folding of
headers in such a way that we can construct a backward compatibility policy
that is 100% compatible with the 3.2 API, while allowing a new policy to
implement the email6 API.  (2) it adds that backward compatibility policy and
refactors the test suite so that the only differences between the 3.2
test_email.py file and the 3.3 test_email.py file is some small changes in
test framework and the addition of tests for bugs fixed that apply to the 3.2
API.

There are some additional teaks, such as moving just the code needed for the
compatibility policy into _policybase, so that the library code can import
only _policybase.  That way the new code that will be added for email6
will only get imported when a non-compatibility policy is imported.

This protects from possibilty of gc running just after fork.

Patch by Charles-François Natali

Fix suggested by Itay Brandes

Patch by Victor Stinner.

Patch by Victor Stinner.

Previously, if you passed in a bytes object, it would create a whole
new object.

There's no bufsize argument anymore, and os.fdopen() is only a very
thin wrapper around open() anymore. Change the documentation to
reflect that.