digits of precision for floats.

Addresses the float -> string conversion, using David Gay's code which
was added in Mark Dickinson's checkin r71663.

Also addresses these, which are intertwined with the short repr
changes:

- Issue #5772: format(1e100, '<') produces '1e+100', not '1.0e+100'
- Issue #5515: 'n' formatting with commas no longer works poorly
    with leading zeros.
- PEP 378 Format Specifier for Thousands Separator: implemented
    for floats.

 - incorporate and adapt David Gay's dtoa and strtod
   into the Python core
 - on platforms where we can use Gay's code (almost
   all!), repr(float) is based on the shortest
   sequence of decimal digits that rounds correctly.
 - add sys.float_repr_style attribute to indicate
   whether we're using Gay's code or not
 - add autoconf magic to detect and enable SSE2
   instructions on x86/gcc
 - slight change to repr and str:  repr switches
   to exponential notation at 1e16 instead of
   1e17, str switches at 1e11 instead of 1e12

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r71640 | thomas.heller | 2009-04-16 08:26:33 +0200 (Do, 16 Apr 2009) | 1 line

  Remove unneeded code.
........
  r71641 | thomas.heller | 2009-04-16 08:42:02 +0200 (Do, 16 Apr 2009) | 2 lines

  Remove duplicated function declaration.
  Make _pagesize static.
........

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r71651 | r.david.murray | 2009-04-16 14:12:53 -0400 (Thu, 16 Apr 2009) | 2 lines

  Less red ink (warning->note) and add link to def of side-by-side assembly.
........

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r71631 | r.david.murray | 2009-04-15 18:33:07 -0400 (Wed, 15 Apr 2009) | 4 lines

  Fix for issue3440: add warning to subprocess discussion of
  env parameter that on Windows SystemRoot is required in order
  to run side-by-side assemblies.
........

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r71627 | benjamin.peterson | 2009-04-15 16:26:36 -0500 (Wed, 15 Apr 2009) | 4 lines

  call __float__ on str subclasses #5759

  tests by R. David Murray
........

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r71589 | tarek.ziade | 2009-04-13 22:14:54 +0200 (Mon, 13 Apr 2009) | 1 line

  pep8-fied
........

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r71585 | tarek.ziade | 2009-04-13 22:03:44 +0200 (Mon, 13 Apr 2009) | 1 line

  improved test coverage for distutils.cmd
........

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r71579 | r.david.murray | 2009-04-13 12:56:32 -0400 (Mon, 13 Apr 2009) | 2 lines

  Add missing NEWS item for issue1161031 fix.
........

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r71569 | tarek.ziade | 2009-04-13 14:42:26 +0200 (Mon, 13 Apr 2009) | 1 line

  deactivate test_search_cpp under win32
........

........
  r70958 | kristjan.jonsson | 2009-04-01 16:08:34 +0000 (Mi, 01 Apr 2009) | 3 lines

  http://bugs.python.org/issue5623
  Dynamically discoverd the size of the ioinfo struct used by the crt for its file descriptors.  This should work across all flavors of the CRT.  Thanks to Amaury Forgeot d'Arc
  Needs porting to 3.1
........

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r71560 | tarek.ziade | 2009-04-13 14:34:01 +0200 (Mon, 13 Apr 2009) | 1 line

  Fixed #5607: Distutils test_get_platform was failing fo Mac OS X fat binaries.
........

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r71553 | r.david.murray | 2009-04-12 21:06:46 -0400 (Sun, 12 Apr 2009) | 3 lines

  Adjust test_asyncore to account for intentional asyncore behavior change
  introduced by r70934 that was causing a test failure when run under -O.
........

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r71548 | eric.smith | 2009-04-12 20:29:50 -0400 (Sun, 12 Apr 2009) | 1 line

  Fixed incorrect object passed into format_float_internal(). This was resulting in a conversion being done twice.
........

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r71539 | benjamin.peterson | 2009-04-12 15:24:56 -0500 (Sun, 12 Apr 2009) | 1 line

  remove useless import
........

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r71533 | tarek.ziade | 2009-04-12 19:02:08 +0200 (Sun, 12 Apr 2009) | 1 line

  removed string usage and added a test for _clean
........

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r71528 | tarek.ziade | 2009-04-12 18:45:32 +0200 (Sun, 12 Apr 2009) | 1 line

  added a test for finalize_options
........

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r71523 | tarek.ziade | 2009-04-12 18:31:24 +0200 (Sun, 12 Apr 2009) | 1 line

  added a simple test for search_cpp
........

string.maketrans() which actually works on bytes.  (Also closes #5675.)

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r70873 | josiah.carlson | 2009-03-31 15:32:34 -0400 (Tue, 31 Mar 2009) | 2 lines

  This resolves issue 1161031.  Tests pass.
........
  r70904 | josiah.carlson | 2009-03-31 17:49:36 -0400 (Tue, 31 Mar 2009) | 3 lines

  Made handle_expt_event() be called last, so that we don't accidentally read
  after closing the socket.
........
  r70934 | josiah.carlson | 2009-03-31 21:28:11 -0400 (Tue, 31 Mar 2009) | 2 lines

  Fix for failing asyncore tests.
........
  r71490 | r.david.murray | 2009-04-11 13:52:56 -0400 (Sat, 11 Apr 2009) | 4 lines

  Make test_asyncore tests match code changes introduced by the
  fix to Issue1161031, refactoring the test to simplify it in
  the process.
........

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r71513 | tarek.ziade | 2009-04-12 17:03:50 +0200 (Sun, 12 Apr 2009) | 1 line

  pep8-fied the module before adding tests
........

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r71509 | tarek.ziade | 2009-04-12 16:53:51 +0200 (Sun, 12 Apr 2009) | 1 line

  removed the print statements and added a test
........

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r71024 | georg.brandl | 2009-04-02 04:47:44 +0200 (Do, 02 Apr 2009) | 4 lines

  In PyErr_GivenExceptionMatches, temporarily bump the recursion
  limit, so that in the most common case PyObject_IsSubclass will
  not raise a recursion error we have to ignore anyway.
........
  r71058 | georg.brandl | 2009-04-02 20:09:04 +0200 (Do, 02 Apr 2009) | 3 lines

  PyErr_NormalizeException may not set an error, so convert the PyErr_SetObject
  call on hitting the recursion limit into just assigning it to the arguments provided.
........