Prep 5.5.0 release.

Avoid most calls to time.sleep() in the tests by mocking time.[gm]time

grammar (typo)

Lifted from PR #164.

This happens on Python 3.7, evidinced by Appveyor (https://ci.appveyor.com/project/jimfulton/zodb/build/1.0.176/job/ehutnhnvbm28w5xl) but not travis because we run coverage on travis, and coverage disables refcounts of None.

This also makes us more deterministic. (Based on code I wrote for nti.testing.)

On my machine, this speeds up a sequential test run from 4:08 to 1:30.

We no longer get much benefit from running the layer tests in
parallel, in fact it looks like the overhead actually slows them down
slightly, so stop doing that.

p64 and p64 are faster and create more informative exceptions.

Fixes #216.

Add Python 3.7 support. 

And fix all the deprecations it reports.

Fixes #214 and fixes #215 (well there's one remaining deprecation in 'manuel')

Based on #217 so that all tests pass.

Remove support for 'python setup.py test'. [skip ci]

Fix tests with, and depend on, persistent 4.4.

Skipping CI because tests will fail until #217 is merged.

Fixes #218.

Fixes #213.

This is required to avoid a memory leak on Python 2.7.

See #203.

prefect -> prefetch

(Previously, protocol 2 was used for Python 2.)

  The zodbpickle package provides a `zodbpickle.binary` string type
  that should be used in Python 2 to cause binary strings to be saved
  in a pickle binary format, so they can be loaded correctly in
  Python 3.  Pickle protocol 3 is needed for this to work correctly.

- Object identifiers in persistent references are saved as
  `zodbpickle.binary` strings in Python 2, so that they are loaded
  correctly in Python 3.

Fix typo in doctest documentation of find_global method

* Don't seek on imports

(other than the possible seek for custom importers)

We were seeking to handle blob markers. This has two major drawbacks:

1. It wasn't possible to use a non-seekable file.  A use case for
   export/import is to copy database data.  An intermediate file, and
   associated I/O, could be avoided using a pipe, but pipes aren't
   seekable.

2. Seeks cause file-buffer data to be discarded, making IO far more
   expensive.

We didn't really need blob markers, because the preceeding blob data
records serve as markers.  (Now we're stuck with them for backward
compatibility.)

* Make cp's buffer size larger and configurable.

* Use the storage temprary directory when importing blobs

To avoid an extra copy.

Also, allow the copy (cp) buffer sie to be overridden on export.

(I see no obvious way to plumb it on import. :( )

* Oops, need to use a binary literal (Python 3)

* Respond to PR comments

Especially with regard to write capacity and search.

This hasn't been updated in a while and I feel that it is out of date.

I feel that this could use more work, but I'm anxious to do an
incremental improvement given the screencast I just released. :)

Report full oid of missing object during pack

Review Connection/DB closure

In addition to some micro-optimisation, this fixes the following minor issues:
- Closing a DB left Connections registered to the global transaction manager.
  Which broke at least multi-db because only such primary connections were
  ignored with monkey-patches.
- Stop calling the real newTransaction when aborting an existing transaction.
  This was not the case if explicit_transactions=0.

Unpin specific PyPy versions, blacklist 3.3 in `setup.py`

* Use a higher pickle protocol (2) for serializing objects on Python 2

Previously protocol 1 was used. This is more efficient for new-style
classes (all persistent objects are new-style), according to the docs,
at the cost of being very slightly less space efficient for old-style
classes.

In tests of a persistent object with two trivial numeric attributes,
the higher protocol was 12 bytes smaller, and serialized and
deserialized 1us faster. Introducing a reference to another new-style
class for a more realistic test made the higher protocol twice as fast
to serialize (20.5 vs 10.3us), almost half the size (215 vs 142
bytes), and it deserialized 30% faster (6.5 vs 4.6us).

On Python 2, this will now allow open ``file`` objects to be
pickled (loading the object will result in a closed file); previously
this would result in a ``TypeError`` (as does under Python 3). We had
tests that you couldn't do that with a BlobFile so I had to update it
to still make that true.

I wouldn't recommend serializing arbitrary open files under Python
2 (for one thing, they can't trivially be deserialized in Python 3),
but I didn't take any steps to prevent it either. Since this hasn't
been possible, there shouldn't be code in the wild that is trying to
do it---and it wouldn't be forward compatible with Python 3 either.