Follow up to #22642.

🎉 🍻

Still more to do. Please assist!

We know the query will return exactly one row for each entry in the
`ids` array, so we can do all the limit/offset calculations on that
array, in advance.

I also split our new ordered-ids behaviour out of the existing
`find_some` method: especially with this change, the conditionals were
overwhelming the actual logic.

Since the attributes API is new in Rails 5, we don't actually need to keep
the behavior of `attribute :point`, as it's not a breaking change.

Support passing the schema name as a prefix to table name in
`ConnectionAdapters::SchemaStatements#indexes`. Previously the prefix would
be considered a full part of the index name, and only the schema in the
current search path would be considered.

It was changed by mistake at 428d47ad

This reverts commit 4d06ea9a, reversing
changes made to e9d15072.

Reason: This will break oracle-enhanced, see
https://github.com/rsim/oracle-enhanced/blob/3c42131db82b64ac41645db3affc6e4650289df6/lib/active_record/connection_adapters/oracle_enhanced_adapter.rb#L1254

All public methods are tasks, so we need to move it to protected
visibility.

It's pretty common for folks to monkey patch `ActiveRecord::Base` to
work around an issue or introduce extra functionality. Instead of
shoving even more stuff in `ActiveRecord::Base`, `ApplicationRecord` can
hold all those custom work the apps may need.

Now, we don't wanna encourage all of the application models to inherit
from `ActiveRecord::Base`, but we can encourage all the models that do,
to inherit from `ApplicationRecord`.

Newly generated applications have `app/models/application_record.rb`
present by default. The model generators are smart enough to recognize
that newly generated models have to inherit from `ApplicationRecord`,
but only if it's present.

Fix white-space

Add test case demonstrating flags are received by the adapter

```
# before
DEPRECATION WARNING:               Time columns will become time zone aware in Rails 5.1. This
              still causes `String`s to be parsed as if they were in `Time.zone`,
              and `Time`s to be converted to `Time.zone`.

              To keep the old behavior, you must add the following to your initializer:

                  config.active_record.time_zone_aware_types = [:datetime]

              To silence this deprecation warning, add the following:

                  config.active_record.time_zone_aware_types << :time
```

```
# after
DEPRECATION WARNING: Time columns will become time zone aware in Rails 5.1. This
still causes `String`s to be parsed as if they were in `Time.zone`,
and `Time`s to be converted to `Time.zone`.

To keep the old behavior, you must add the following to your initializer:

    config.active_record.time_zone_aware_types = [:datetime]

To silence this deprecation warning, add the following:

    config.active_record.time_zone_aware_types << :time
```

We currently generate an unbounded number of prepared statements when
`limit` or `offset` are called with a dynamic argument. This changes
`LIMIT` and `OFFSET` to use bind params, eliminating the problem.

`Type::Value#hash` needed to be implemented, as it turns out we busted
the query cache if the type object used wasn't exactly the same object.

This drops support for passing an `Arel::Nodes::SqlLiteral` to `limit`.
Doing this relied on AR internals, and was never officially supported
usage.

Fixes #22250.

Some backends allow `LIMIT 1,2` as a shorthand for `LIMIT 1 OFFSET 2`.
Supporting this in Active Record massively complicates using bind
parameters for limit and offset, and it's trivially easy to build an
invalid SQL query by also calling `offset` on the same `Relation`.

This is a niche syntax that is only supported by a few adapters, and can
be trivially worked around by calling offset explicitly.

Even though this means more things to change when we bump after a
release, it's more important that our examples are directly copyable.

.. it also showed a deprecation warning, but we obviously needn't retain
that.

If we use a real version, at best that'll be an onerous update required
for each release; at worst, it will encourage users to write new
migrations against an older version than they're using.

The other option would be to leave these bare, without any version
specifier. But as that's just a variant spelling of "4.2", it would seem
to raise the same concerns as above.

This reverts commit 6d5b1fdf.

`eager_load` and `references` can include hashes, which won't match up
with `references`

A test case has been added to demonstrate the problem

This was changed in 421c81bd, as `exists?` blows up if you are eager
loading a polymorphic association, as it'll try to construct a join to
that table. The previous change decided to execute a `count` instead,
which wouldn't join.

Of course, the only time we actually need to perform a join on the eager
loaded values (which would perform a left outer join) is if they're
being referenced in the where clause. This doesn't affect inner joins.

We currently generate an unbounded number of prepared statements when
`limit` or `offset` are called with a dynamic argument. This changes
`LIMIT` and `OFFSET` to use bind params, eliminating the problem.

`Type::Value#hash` needed to be implemented, as it turns out we busted
the query cache if the type object used wasn't exactly the same object.

This drops support for passing an `Arel::Nodes::SqlLiteral` to `limit`.
Doing this relied on AR internals, and was never officially supported
usage.

Fixes #22250.

Some backends allow `LIMIT 1,2` as a shorthand for `LIMIT 1 OFFSET 2`.
Supporting this in Active Record massively complicates using bind
parameters for limit and offset, and it's trivially easy to build an
invalid SQL query by also calling `offset` on the same `Relation`.

This is a niche syntax that is only supported by a few adapters, and can
be trivially worked around by calling offset explicitly.

This reverts commit c2e70ca9, reversing
changes made to b0e5fc27.

this broke the build

In the doc the `dependent` option was set with: `dependent: destroy`.
This is not working because destroy would call the method of the activerecord::base object.

The right way is: `dependent: :destroy`

The problem was that when saving an object, we would
call touch_later on the parent which wont be saved immediteally, and
it wont call any callbacks. That was working one level up because
we were calling touch, during the touch_later commit phase. However that still
didnt solve the problem when you have a 3+ levels of parents to be touched,
as calling touch would affect the parent, but it would be too late to run callbacks
on its grand-parent.

The solution for this, is instead, call touch_later upwards when the first
touch_later is called. So we make sure all the timestamps are updated without relying
on callbacks.

This also removed the hard dependency BelongsTo builder had with the TouchLater module.
So we can still have the old behaviour if TouchLater module is not included.

[fixes 5f5e6d92]
[related #19324]

Those are actually shortcuts for `after_commit`.

Before:

    after_commit :add_to_index_later, on: :create
    after_commit :update_in_index_later, on: :update
    after_commit :remove_from_index_later, on: :destroy

After:

    after_create_commit  :add_to_index_later
    after_update_commit  :update_in_index_later
    after_destroy_commit :remove_from_index_later

The `rake db:test:*` tasks were deprecated in #13528, but were
undeprecated and added back in via #17739.