for memoization on a class that uses slots.
- apply many more __slots__.  mem use for nova now at 46% savings

to start up listener collections; this pulls the overhead off of construction
and makes performance much like the descriptor version, while still allowing
slots.   Fix up some profiles.

allowing us to move to __slots__

size of the many per-column objects we're hitting, but somehow the overall memory is
hardly being reduced at all in initial testing

as that of :ticket:`3199`, when the ``named=True`` parameter
would be used.  Some events would fail to register, and others
would not invoke the event arguments correctly, generally in the
case of when an event was "wrapped" for adaption in some other way.
The "named" mechanics have been rearranged to not interfere with
the argument signature expected by internal wrapper functions.
fixes #3197

ORM events but also engine events, where the usual logic of
"de duplicating" a redundant call to :func:`.event.listen`
with the same arguments would fail, for those events where the
listener function is wrapped.  An assertion would be hit within
registry.py.  This assertion has now been integrated into the
deduplication check, with the added bonus of a simpler means
of checking deduplication across the board.
fixes #3199

is being run itself, either from inside the listener or from a
concurrent thread, now raises a RuntimeError, as the collection used is
now an instance of ``colletions.deque()`` and does not support changes
while being iterated.  Previously, a plain Python list was used where
removal from inside the event itself would produce silent failures.
fixes #3163

sqlalchemy/orm, sqlalchemy/event, sqlalchemy/testing

to get all flake8 passing

event.listen and event.listen_for have a kwarg once added in 0.9.4 (not 0.9.3) CHANGELOG agrees with this as well. (as does my manual testing)