* Depend on websocket-driver >= 0.6.1

Added a test around `NO_AUTO_VALUE_ON_ZERO`:

- Add set_locale to detect suitable locale
- Make feature compatible with Rails 5.x

* Support hash as first argument for `assert_difference`.

This allows to specify multiple numeric differences in the same assertion.
Example:

    assert_difference 'Article.count' => 1, 'Notification.count' => 2 do
      # post :create, params: { article: {...} }
    end

* Support error message when passing a hash as a first parameter

* Format CHANGELOG properly

[Julien Meichelbeck + Rafael Mendonça França]

Support for connection pooling on mem cache store

Added a note about bootsnap in the Rails 5.2 upgrade guide [ci skip]

Allow `false` for `config.generators.system_tests=`

down is only called with a block from the rake tasks where it passes a
`SCOPE`. Technically this was tested but since we don't run all the
migrations we're not actually testing the down works with a `SCOPE`. To
ensure we're testing both we can run `db:migrate` again to migrate users
and then run `down` with a scope to test that only the bukkits migration
is reverted.

Updates test to prevent having to fix regressions like we did in
4d4db4c8.

This slipped in as part of a2827ec9.

The UJS build has been failing with Chrome failed to start. This commit
fixes it by adding the option `--no-sandbox`. Travis removed the sanbox
option which is why Chrome crashes.

Ref https://github.com/travis-ci/travis-ci/issues/8836

Example failure: https://travis-ci.org/rails/rails/jobs/330396750

This reverts commit edc54fd2, reversing
changes made to a5922f13.

As discussed, this is not an appropriate place to make assumptions about
ARGV, or to write to stdout: config/boot.rb is a library and is required
by other applictions, with which we have no right to interfere.

Merge pull request #31727 from eileencodes/refactor-migration-classes-to-allow-for-migrations_paths-on-conn

Refactor migration to move migrations paths to connection

Rails has some support for multiple databases but it can be hard to
handle migrations with those. The easiest way to implement multiple
databases is to contain migrations into their own folder ("db/migrate"
for the primary db and "db/seconddb_migrate" for the second db). Without
this you would need to write code that allowed you to switch connections
in migrations. I can tell you from experience that is not a fun way to
implement multiple databases.

This refactoring is a pre-requisite for implementing other features
related to parallel testing and improved handling for multiple
databases.

The refactoring here moves the class methods from the `Migrator` class
into it's own new class `MigrationContext`. The goal was to move the
`migrations_paths` method off of the `Migrator` class and onto the
connection. This allows users to do the following in their
`database.yml`:

```
development:
  adapter: mysql2
  username: root
  password:

development_seconddb:
  adapter: mysql2
  username: root
  password:
  migrations_paths: "db/second_db_migrate"
```

Migrations for the `seconddb` can now be store in the
`db/second_db_migrate` directory. Migrations for the primary database
are stored in `db/migrate`".

The refactoring here drastically reduces the internal API for migrations
since we don't need to pass `migrations_paths` around to every single
method. Additionally this change does not require any Rails applications
to make changes unless they want to use the new public API. All of the
class methods from the `Migrator` class were `nodoc`'d except for the
`migrations_paths` and `migrations_path` getter/setters respectively.

Mention `config.generators.system_tests` in
the "Configuring Rails Applications" guide.

- The mysql `NO_AUTO_VALUE_ON_ZERO` mode should be disabled when inserting fixtures in bulk, this PR adds a test to make sure we don't remove it by mistake
- If we live this mode enabled, a statement like this wouldn't work and a `Duplicate entry '0' for key 'PRIMARY'` error will be raised. That's because `DEFAULT` on auto_increment will return 0

```sql
INSERT INTO `aircraft` (`id`, `name`, `wheels_count`) VALUES (DEFAULT, 'first', 2), (DEFAULT, 'second', 3)
```

Move browser config to its own class

Convert keys to binary in the Redis cache store

Update "Layouts and Rendering in Rails" guide [ci skip]

  - Remove mention about `ActionView::TemplateHandlers` since this
    module was removed by c1304098.
    Change word `subclasses` to `nested classes`.
    See c7408a0e
  - Remove useless sentence "Beginning with Rails 2, the standard extensions
    are `.erb` for ERB (HTML with embedded Ruby), and `.builder` for Builder (XML generator)."

Fix encoding errors when using the pure-Ruby Redis driver instead of Hiredis. Dodge incompatibilities between UTF-8 and arbitrary value encodings, which rear their heads when the Redis driver tries to build a single command string from incompatibly-encoded keys and values.

Update the long key handling test so it triggers truncation in the Redis cache store.

Fix relation merger issue with `left_outer_joins`.

Currently `object.save` will unfreeze the object, due to
`changes_applied` replaces frozen `@attributes` to new `@attributes`.

Since originally destroyed objects are not allowed to be mutated, `save`
and `save!` should not return success in that case.

Fixes #28563.

If `collation` is given without `charset`, it may generate invalid SQL.
For example `create_database(:matt_aimonetti, collation: "utf8mb4_bin")`:

```
> CREATE DATABASE `matt_aimonetti` DEFAULT CHARACTER SET `utf8` COLLATE `utf8mb4_bin`;
ERROR 1253 (42000): COLLATION 'utf8mb4_bin' is not valid for CHARACTER SET 'utf8'
```

In MySQL, charset is used to find the default collation. If `collation`
is given explicitly, it is not necessary to give extra charset.

Allow unscoping of left_outer_joins

Change test case name to FormatValidationTest

I believe that this was caused by a copy/paste mistake.

Because `Relation` already have Arel `table`.

This is a complete fix to #30995.

Originally alias tracker will only track table aliases on
`Arel::Nodes::Join`, other args are ignored.

Since c5ab6e51, parent aliases hash will be passed then it caused the
regression #30995.

It is enough to pass list of `Arel::Nodes::Join` simply, not need to
pass garbage args which will be ignored.

Fix comment about initializers to adapt to the fact

Use SHA-1 for non-sensitive digests by default

Fix doc typo [ci-skip]