This feature was supposed to be part of the initial email6 checkin, but it got
lost in my big refactoring.

In this patch I'm not providing an easy way to turn off the errors, but they
only happen when a header is added programmatically, and it is almost never
the right thing to do to allow the duplicate to be added.  An application that
needs to add duplicates of unique headers can create a policy subclass to
allow it.

 * Formatting string, int, float and complex use the _PyUnicodeWriter API. It
   avoids a temporary buffer in most cases.
 * Add _PyUnicodeWriter_WriteStr() to restore the PyAccu optimization: just
   keep a reference to the string if the output is only composed of one string
 * Disable overallocation when formatting the last argument of str%args and
   str.format(args)
 * Overallocation allocates at least 100 characters: add min_length attribute
   to the _PyUnicodeWriter structure
 * Add new private functions: _PyUnicode_FastCopyCharacters(),
   _PyUnicode_FastFill() and _PyUnicode_FromASCII()

The speed up is around 20% in average.

Previously a Python 2.x compatible hack was used for
multiprocessing.sharedctypes.Array().  Also the documented
signature was wrong.

Proxy classes in multiprocessing do not need these methods in Python 3.x.

Patch by Sidney San Martín.

Add attrib keyword to Element and SubElement in _elementtree.
Patch developed with Ezio Melotti.

Issue #14775: Fix a potential quadratic dict build-up due to the garbage collector repeatedly trying to untrack dicts.
Additional comments by Tim Silk.

Issue #14775: Fix a potential quadratic dict build-up due to the garbage collector repeatedly trying to untrack dicts.
Additional comments by Tim Silk.

An issue in ctypes.c_longdouble, ctypes.c_double, and ctypes.c_float that
caused an incorrect exception to be returned in the case of overflow has been
fixed.

An issue in ctypes.c_longdouble, ctypes.c_double, and ctypes.c_float that
caused an incorrect exception to be returned in the case of overflow has been
fixed.

Added python3 symlink; removed Distribute-related code, docs and comments; changed Mac OS X computation to determine framework builds.

Patch by Roger Serwy.

This commit also restores the news item for 167256 that it looks like
Terry inadvertently deleted.  (Either that, or I don't understand
now merging works...which is equally possible.)

Original patch by Roger Serwy.

Which also means that it is now producing *something* for any base64
payload, which is what leads to the couple of older test changes in
test_email.  This is a slightly backward incompatible behavior change,
but the new behavior is so much more useful than the old (you can now
*reliably* detect errors, and any program that was detecting errors by
sniffing for a base64 return from get_payload(decode=True) and then doing
its own error-recovery decode will just get the error-recovery decode
right away).  So this seems to me to be worth the small risk inherent
in this behavior change.

This patch also refactors the defect tests into a separate test file,
since they are no longer just parser tests.

This patch also deprecates the MalformedHeaderDefect.  My best guess is that
this defect was rendered obsolete by a refactoring of the parser, and the
corresponding defect for the new parser (which this patch introduces) was
overlooked.

Also adds the TOC entry for headerregistry.

When I made the checkin of the provisional email policy, I knew that
Address and Group needed to be made accessible from somewhere.  The more
I looked at it, though, the more it became clear that since this is a
provisional API anyway, there's no good reason to hide headerregistry as
a private API.  It was designed to ultimately be part of the public API,
and so it should be part of the provisional API.

This patch fully documents the headerregistry API, and deletes the
abbreviated version of those docs I had added to the provisional policy
docs.