Fix habtm reflection
Conflicts:
    activerecord/CHANGELOG.md
    activerecord/lib/active_record/counter_cache.rb
    activerecord/lib/active_record/reflection.rb
    activerecord/test/cases/reflection_test.rb

This mirrors the layout of abstract adapter and puts the definitions
inside the `PostgreSQL` namespace (no longer under the adapter namespace).

/cc @kares

It seems that #where! is not designed to be used as a chained where.
See initial implementation at 8c2c6051.
So, no need to check twice.

We should not test #where!

https://github.com/rails/rails/pull/15285#discussion_r13018316

It appears to have been used at some point in the past, but is no longer
used in any meaningful way. Whether a column is considered primary is
a property of the model, not the schema/column. This also removes the
need for yet another layer of caching of the model's schema, and we can
leave that to the schema cache.

The use of default procs was unnessecary, made the code confusing to
follow, and made marshalling needlessly complex.

Should make it a little easier to find the information. Also added note to look below for the definition of the generated methods.

2d73f5ae forces AR to enter the `define_attribute_methods` method whenever it
instantiate a record from the `init_with` entry point. This is a potential
performance hotspot, because `init_with` is called from all `find*` family
methods, and `define_attribute_methods` is slow because it tries to acquire
a lock on the mutex everytime it is entered.

By using [DCL](http://en.wikipedia.org/wiki/Double-checked_locking), we can
avoid grabbing the lock most of the time when the attribute methods are already
defined (the common case). This is made possible by the fact that reading an
instance variable is an atomic operation in Ruby.

Credit goes to Aaron Patterson for pointing me to DCL and filling me in on the
atomicity guarantees in Ruby.

[*Godfrey Chan*, *Aaron Patterson*]

* * *

This bug can be triggered when serializing record R (the instance) of type C
(the class), provided that the following conditions are met:

1. The name of one or more columns/attributes on C/R matches an existing private
   method on C (e.g. those defined by `Kernel`, such as `format`).

2. The attribute methods have not yet been generated on C.

In this case, the matching private methods will be called by the serialization
code (with no arguments) and their return values will be serialized instead. If
the method requires one or more arguments, it will result in an `ArgumentError`.

This regression is introduced in d1316bb1.

* * *

Attribute methods (e.g. `#name` and `#format`, assuming the class has columns
named `name` and `format` in its database table) are lazily defined. Instead of
defining them when a the class is defined (e.g. in the `inherited` hook on
`ActiveRecord::Base`), this operation is deferred until they are first accessed.

The reason behind this is that is defining those methods requires knowing what
columns are defined on the database table, which usually requires a round-trip
to the database. Deferring their definition until the last-possible moment helps
reducing unnessary work, especially in development mode where classes are
redefined and throw away between requests.

Typically, when an attribute is first accessed (e.g. `a_book.format`), it will
fire the `method_missing` hook on the class, which triggers the definition of
the attribute methods. This even works for methods like `format`, because
calling a private method with an explicit receiver will also trigger that hook.

Unfortunately, `read_attribute_for_serialization` is simply an alias to `send`,
which does not respect method visibility. As a result, when serializing a record
with those conflicting attributes, the `method_missing` is not fired, and as a
result the attribute methods are not defined one would expected.

Before d1316bb1, this is negated by the fact that calling the `run_callbacks`
method will also trigger a call to `respond_to?`, which is another trigger point
for the class to define its attribute methods. Therefore, when Active Record
tries to run the `after_find` callbacks, it will also define all the attribute
methods thus masking the problem.

* * *

The proper fix for this problem is probably to restrict `read_attribute_for_serialization`
to call public methods only (i.e. alias `read_attribute_for_serialization` to
`public_send` instead of `send`). This however would be quite risky to change
in a patch release and would probably require a full deprecation cycle.

Another approach would be to override `read_attribute_for_serialization` inside
Active Record to force the definition of attribute methods:

   def read_attribute_for_serialization(attribute)
     self.class.define_attribute_methods
     send(attribute)
   end

Unfortunately, this is quite likely going to cause a performance degradation.

This patch therefore restores the behaviour from the 4-0-stable branch by
explicitly forcing the class to define its attribute methods in a similar spot
(when records are initialized). This should not cause any extra roundtrips to
the database because the `@columns` should already be cached on the class.

Fixes #15188.

Columns and injected types no longer have any conditionals based on the
format of SQL type strings! Hooray!

Ideally types will be usable without having to specify a sql type
string, so we should keep the information related to parsing them on the
adapter or another object.

We're going to want all of the benefits of the type map object for
registrations, including block registration and real aliasing. Moves
type name registrations to the adapter, and aliases the OIDs to the
named types

Determining things like precision and scale in postgresql will require
the given blocks to take additional arguments besides the OID.

- Adds the ability to handle additional arguments to `TypeMap`
- Passes the column type to blocks when looking up PG types

The only type that has a scale is decimal. There's a special case where
decimal columns with 0 scale are type cast to integers if the scale is
not specified. Appears to only affect schema dumping.

This reverts commit 9a1abedc, reversing
changes made to c72d6c91.

Conflicts:
	activerecord/CHANGELOG.md
	activerecord/test/models/comment.rb

This change break integration with activerecord-deprecated_finders so
I'm reverting until we find a way to make it work with this gem.

As we promote these classes to first class concepts, these classes are
starting to gain enough behavior to warrant being moved into their own
files. Many of them will become quite large as we move additional
behavior to the type objects.

  - `extract_precision`, `extract_limit`, and `extract_default` probably need to follow.
  - would be good to remove the delegation `Column#extract_scale`.

/cc @sgrif

The foreign_key could be `String` and just doing `owners_map[owner_key]`
could return `nil`.
To prevent this bug, we should `to_s` both keys if their types are
different.

Fixes #14734.

This reverts commit e94e6c27.

Conflicts:
	activerecord/CHANGELOG.md

The original commit was reverted only to be safe since #14407 were reported.
We don't have any proof we added a regression with the original commit
so reverting it now will give us more problem.

Closes #14407

All subclasses of column were now delegating `type_cast` to their
injected type object. We can remove the overriding methods, and
generalize it on the `Column` class itself. This also enabled us to
remove several column classes completely, as they no longer had any
meaningful behavior of their own.