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

sort_tables_and_constraints function.
- The DDL generation system of :meth:`.MetaData.create_all`
and :meth:`.Metadata.drop_all` has been enhanced to in most
cases automatically handle the case of mutually dependent
foreign key constraints; the need for the
:paramref:`.ForeignKeyConstraint.use_alter` flag is greatly
reduced.  The system also works for constraints which aren't given
a name up front; only in the case of DROP is a name required for
at least one of the constraints involved in the cycle.
fixes #3282

to complement the :attr:`.Table.foreign_keys` collection,
as well as :attr:`.ForeignKeyConstraint.referred_table`.

a type that was also a :class:`.TypeDecorator` would fail with
Python's "Cannot create a consistent method resolution order (MRO)"
error, when any kind of SQL comparison expression were used against
an object using this type.

- fix "temporary_tables" requirement

that quoting should be applied in %(name)s under with_hint.

to the aliasing syntax, as well as a new CTE feature
:meth:`.CTE.suffix_with`, which is useful for adding in special
Oracle-specific directives to the CTE.
fixes #3220

return type is not strictly assumed to be boolean; it now
returns a :class:`.Boolean` subclass called :class:`.MatchType`.
The type will still produce boolean behavior when used in Python
expressions, however the dialect can override its behavior at
result time.  In the case of MySQL, while the MATCH operator
is typically used in a boolean context within an expression,
if one actually queries for the value of a match expression, a
floating point value is returned; this value is not compatible
with SQLAlchemy's C-based boolean processor, so MySQL's result-set
behavior now follows that of the :class:`.Float` type.
A new operator object ``notmatch_op`` is also added to better allow
dialects to define the negation of a match operation.
fixes #3263