Collector #1490 - Added a new zope.conf option to control the character set used to encode unicode data that reaches ZPublisher without any specified encoding.

Putative fix for Collector 1912.

File._read_data():  This uses savepoints just to get oids assigned
and to free memory, with no possibility of rollback.  Therefore
"optimistic" savepoints should be used:  there's no reason here to
insist that any data managers involved support rollback.

The '@' character is now allowed in object ids (RFC 1738 allows it).
Expanded tests for _checkId.

- some cleanup

with DirectoryStorage

  the TransientObject's __repr__ method.

test causes a Unauthorized exception to be raised, which now correctly is *not*
handled by the publisher, as requested.

**This change needs to be propagated to the Five repository to become
permanent!**

Prevent 'render' function in Expressions.py for being called for basic types (minor speed improvement).  See  http://www.zope.org/Collectors/Zope/1890. 

        ac_aquire() ignoring the default argument

Collector #1864, #1906: fixed header normalization in appendHeader()

      - Collector #1899: fixed migration issue when using export/import for
        ZCatalog instances

packages than single pieces

  with Python
- remove use of the long-deprecated xmllib module, again using the Expat
  that comes with Python as a replacement

use it

In the interest of keeping myself sane, I've left the "Trunk only
(unreleased)" section alone