Fix Gemfile.lock

- Resque version was locked to < 1.26 in 92f869a0 but
  Gemfile.lock was not updated.

regenerate application_mailer.rb if it is missing

The BINARY Operator is only needed for string columns

Right now referencing the constant `AbstractController::Rendering`
causes `ActionView::Base` to be loaded, and thus the load hooks for
action_view are run. If that load hook references any part of action
view that then references action controller (such as
`ActionView::TestCase`), the constant `AbstractController::Rendering`
will attempt to be autoloaded and blow up.

With this change, `ActionView::LoadPaths` no longer requires
`ActionView::Base` (which it had no reason to require). There was a
needed class from `AbstractController::Base` in the Rendering module,
which I've moved into its own file so we don't need to load
all of `AbstractController::Base` there.

This commit fixes
https://github.com/rails/rails-controller-testing/issues/21

Clarify has_many :dependent option docs [ci skip]

Clarifies the documentation here to mean all options are for when the
relation is destroyed; also now reflects the documentation on this same
option found in the has_one section.

There was some subtle breakage caused by #18774, when we removed
`#original_exception` in favor of `#cause`. However, `#cause` is
automatically set by Ruby when raising an exception from a rescue block.
With this change, we will use whichever handler has the highest priority
(whichever call to `rescue_from` came last). In cases where the outer
has lower precidence than the cause, but the outer is what should be
handled, cause will need to be explicitly unset.

Fixes #23925

I'm unsure how this passed CI in the pull request.

Default rendering behavior if respond_to collector doesn't have a block.

Prevent `Marshal.load` from looping infinitely

Fix `thread_mattr_accessor` thread-local variable naming

Extract `default_primary_key?` to refactor `column_spec_for_primary_key`

Dump `bigint` instead of `integer` with `limit: 8` for schema dumper

Passing `table_name` to `Column#initialize` to avoid `instance_variable_set`

use `ActiveSupport::Reloader.to_prepare` instead of deprecated `ActionDispatch::Reloader.to_prepare` [ci skip]

use `ActiveSupport::Reloader.to_prepare` instead of deprecated `ActionDispatch::Reloader.to_prepare` [ci skip]

The current implentation of `thread_mattr_accessor` is setting
differently-named thread variables when defining class and
instance writer methods, so the method isn't working as documented:

    Account.user = "DHH"
    Account.user # => "DHH"
    Account.new.user # => nil
    a = Account.new
    a.user = "ABC" # => "ABC"
    a.class.user # => "DHH"

At this point `:attr_Account_user` and `:attr_Class_user` thread-local
variables have been created. Modify the reader and writer methods to use
the class name instead of 'Class'.

Fix a bug in `Marshal.load` that caused it to loop indefinitely when
trying to autoload a constant that resolved to a different name.

This could occur when marshalling an ActiveRecord 4.0 object (e.g. into
memcached) and then trying to unmarshal it with Rails 4.2. The
marshalled payload contains a reference to
`ActiveRecord::ConnectionAdapters::Mysql2Adapter::Column`, which in
Rails 4.2 resolves to
`ActiveRecord::ConnectionAdapters::AbstractMysqlAdapter::Column`.

Before:

```ruby
create_table "big_numbers", force: :cascade do |t|
  t.integer "bigint_column", limit: 8
end
```

After:

```ruby
create_table "big_numbers", force: :cascade do |t|
  t.bigint "bigint_column"
end
```

Fix bigserial appears with limit 8 for schema dumper

Rename dependencies.rake to cache_digests.rake

Remove unused try require in ActiveSupport::TimeZone

If we're gonna do this right, it will look mighty different from this anyway.
(Looking at you, Rails 5.1).

It isn't being used in any code as of now, so yanking is the best option.

Version 3.1.11 works on Windows again

This undoes https://github.com/rails/rails/commit/7241498e51120b9847a8bc16cf48551db0f3e216

https://github.com/codahale/bcrypt-ruby/issues/128 is fixed and closed.

remove obsolete i18n links from guides [ci skip]

Reorder paragraphs in the Routing API docs

The previous order made sense [when `match` was used twice to point to two different actions](https://github.com/rails/rails/commit/7305ef842b675bf965f063de681a96294577fb84). In this case the note was misleading as posting to `/posts/:id` would still route to `show` action.

Remove needless `break;` [ci skip]

Fixed comments of add_foreign_key method