Would be nice to fix Oracle here

- ensure that kwargs can be modified in-place within InstanceEvents.init
and that these take effect for the __init__ method.
- improve documentation for these and related events, including
that kwargs can be modified in-place.

such as :meth:`.Query.union` now handle the case where the embedded
SELECT statements need to be parenthesized due to the fact that they
include LIMIT, OFFSET and/or ORDER BY.   These queries **do not work
on SQLite**, and will fail on that backend as they did before, but
should now work on all other backends.
fixes #2528

to function for a many-to-one relationship.  The loader used an
API to place "None" into the dictionary which no longer actually
writes a value; this is a side effect of :ticket:`3061`.
- remove InstanceState._initialize() totally, it's used nowhere
else and no longer does what it says it does
- fill in fowards-port version ids throughout the changes for 1.0.9

e.g. it never is automatically cast to the type of element in the
structure.  therefore we really don't need the concept of index_map,
it's overkill.   Replace it with a simple callback in Indexable
where each type does its own setup_getitem(), and we're done.
Add a convenience "index_type" attribute to HSTORE and an
"astext_type" attribute to JSON/JSONB..

index_map.   needs generic tests plus migration documentation.
- in particular, the implicit "astext" within cast() is removed.
PG has a built-in cast from json->jsonb and vice versa so we should
stick with requiring explicit if they want astext.cast().

- ARRAY restored, most of JSON, not JSONB yet