First, eliminate the use of the RingWrapper object and always delete
by index. It was only necessary to allow use of ring.remove(). This
had some performance impact, but mostly saves memory.

Second, eliminate the use of `enumerate` in the hot mru() path. This was the big performance win.

Current results:

** concurrency=2 **
"Transaction",                mysql     before
"Add 3000 Objects",             7424     5486
"Update 3000 Objects",          5699     4141
"Read 3000 Warm Objects",       4571     4003
"Read 3000 Cold Objects",       4932     4204
"Read 3000 Hot Objects",       17295     10416
"Read 3000 Steamin' Objects", 346331     168983

Under PyPy, CPython, and Jython, collections.dequeue is a built-in
type. Using it doubles PyPy's performance under `zodbshootout`
(previously, picklecache.mru() was the biggest bottleneck).

Some very minor tweaks to avoid going through the __getattribute__ path so much have large consequences for `zodbshootout`:

  Results show objects written or read per second. Best of 3. (before/after on pypy2.5.1)
  "Transaction",                before        after      cpython2.7
  "Add 3000 Objects",             534         2340        11711
  "Update 3000 Objects",          330         3035        11333
  "Read 3000 Warm Objects",      1168         1941         9086
  "Read 3000 Cold Objects",      1647         4161         9484
  "Read 3000 Hot Objects",       2647        11086        45220
  "Read 3000 Steamin' Objects", 30523       108138      5321032

The python _p_activate does state management like the C version does: set to changed during jar.setstate to prevent objects that set attributes in their __setstate__ from being registered with the jar, and unconditionally mark the object as saved when finished. This fixes problems with BTrees and DB churn, among other things. Test this.

Deleting the jar of a ghosted Python persistent object shouldn't try to activate it. This was an error encountered testing ZEO:

```python
File "//Projects/GithubSources/ZEO/src/ZEO/tests/testZEO.py", line 1566, in ZEO.tests.testZEO.gracefully_handle_abort_while_storing_many_blobs
Failed example:
    c.tpc_abort(t)
Exception raised:
    Traceback (most recent call last):
      File "//bin/pypy/lib-python/2.7/doctest.py", line 1315, in __run
        compileflags, 1) in test.globs
      File "<doctest ZEO.tests.testZEO.gracefully_handle_abort_while_storing_many_blobs[12]>", line 1, in <module>
        c.tpc_abort(t)
      File "//Projects/VirtualEnvs/ZODBpypy/site-packages/ZODB-4.2.0.dev0-py2.7.egg/ZODB/Connection.py", line 754, in tpc_abort
        self._invalidate_creating()
      File "//Projects/VirtualEnvs/ZODBpypy/site-packages/ZODB-4.2.0.dev0-py2.7.egg/ZODB/Connection.py", line 775, in _invalidate_creating
        del o._p_jar
      File "//Projects/VirtualEnvs/ZODBpypy/site-packages/persistent-4.0.10.dev0-py2.7.egg/persistent/persistence.py", line 287, in __delattr__
        object.__delattr__(self, name)
      File "//Projects/VirtualEnvs/ZODBpypy/site-packages/persistent-4.0.10.dev0-py2.7.egg/persistent/persistence.py", line 92, in _del_jar
        self.__setattr__('_Persistent__jar', None)
      File "//Projects/VirtualEnvs/ZODBpypy/site-packages/persistent-4.0.10.dev0-py2.7.egg/persistent/persistence.py", line 249, in __getattribute__
        oga(self, '_p_activate')()
      File "//Projects/VirtualEnvs/ZODBpypy/site-packages/persistent-4.0.10.dev0-py2.7.egg/persistent/persistence.py", line 358, in _p_activate
        jar.setstate(self)
      File "//Projects/VirtualEnvs/ZODBpypy/site-packages/ZODB-4.2.0.dev0-py2.7.egg/ZODB/Connection.py", line 869, in setstate
        self._setstate(obj)
      File "//Projects/VirtualEnvs/ZODBpypy/site-packages/ZODB-4.2.0.dev0-py2.7.egg/ZODB/Connection.py", line 910, in _setstate
        p, serial = self._storage.load(obj._p_oid, '')
      File "//Projects/GithubSources/ZEO/src/ZEO/ClientStorage.py", line 842, in load
        data, tid = self._server.loadEx(oid)
      File "//Projects/GithubSources/ZEO/src/ZEO/ServerStub.py", line 176, in loadEx
        return self.rpc.call("loadEx", oid)
      File "//Projects/GithubSources/ZEO/src/ZEO/zrpc/connection.py", line 781, in call
        raise inst # error raised by server
    POSKeyError: 0x02'')
```

Fix tests for TimeStamp hashcode under 32-bit Pythons.

While we're optimizing, use a set to combine two 'specialness' checks in getattribute into one. This reduces the ZODB test runtime by about another 4-5%

In _p_accessed combine two exception blocks as suggested by @tseaver, plus actually use the oid value we just cached. This knocks another 2 million calls to __getattribute__ out.

Optimizations.  Primarily, this changes _p_accessed to avoid recursively calling back to __getattribute__. When running the ZODB test suite, this changes the number of times that __getattribute__ is called from over 44 MILLION to just 1.3 million (and changes the cumulative time from 38 seconds to 11 seconds).

Secondarily, this potentially also makes a few things safer for Broken objects.

Detect whether we need to do a gc after sweeping the cache. Document why we would need to or not. Add tests to make sure it works (previously only in the ZODB tests.)

Use `ctypes.c_long` to get the native long type on all platforms; fixes 32-bit TimeStamp hashcode

More functional tests from ZODB: test persistent class invalidation, and changing the _p_oid and _p_jar when not in a jar's cache.

Detect the bitness of the current process in order to better match the C TimeStamp hashcode. Fixes zopefoundation/persistent#21

Detect the bitness of the current process in order to better match the C TimeStamp hashcode. Fixes zopefoundation/persistent#21

More on C and ZODB compatibility for Python PickleCache and Persistent. There are some interconnected lifetime issues that this solves for ZODB's cache tests, and what appear to be some genuine bugs in invalidation.