Make it clear that `exists?` can be chained onto a relation

Disallow raw SQL in dangerous AR methods

Compression has long been available, but opt-in and at a 16kB threshold.
It wasn't enabled by default due to CPU cost. Today it's cheap and
typical cache data is eminently compressible, such as HTML or JSON
fragments.

Compression dramatically reduces Memcached/Redis mem usage, which means
the same cache servers can store more data, which means higher hit
rates.

To disable compression, pass `compress: false` to the initializer.

Currently the environment is not loaded in some db tasks.
Therefore, if use encrypted secrets values in `database.yml`,
`read_encrypted_secrets` will not be true, so the value can not be
used correctly.

To fix this, added `environment` as dependency of `load_config`.
It also removes explicit `environment` dependencies that are no longer
needed.

Fixes #30717

* Supports vanilla Redis, hiredis, and Redis::Distributed.
* Supports Memcached-like sharding across Redises with Redis::Distributed.
* Fault tolerant. If the Redis server is unavailable, no exceptions are
  raised. Cache fetches are treated as misses and writes are dropped.
* Local cache. Hot in-memory primary cache within block/middleware scope.
* `read_/write_multi` support for Redis mget/mset. Use Redis::Distributed
  4.0.1+ for distributed mget support.
* `delete_matched` support for Redis KEYS globs.

Cache regexps generated from acronym_regex

Sort mailer previews alphabetically

Closes #31138.

fix bug on added? method

The documentation claims that given values go through "normal AR type
casting and serialization", which to me implies
`serialize(cast(value))`, not just serialization. The docs were changed
to use this wording in #22492. The tests I cited in that PR (which is
the same test modified in this commit), is worded in a way that implies
it should be using `cast` as well.

It's possible that I originally meant "normal type casting" to imply
just the call to `serialize`, but given that `update_all(archived:
params['archived'])` seems to be pretty common, I'm inclined to make
this change as long as no tests are broken from it.

Do not use `Arel.star` when `ignored_columns`

Make clear that the files are not to be run for interpreters.

Fixes #23847.
Fixes #30690.
Closes #23878.

If there are any ignored columns, we will now list out all columns we
want to be returned from the database.

Includes a regression test.

fix rubocop issues

Make sidekiq and resque integration tests work in ci

Since 8f2490b5, the integration test of sidekiq and resque is not working
in CI.
https://travis-ci.org/rails/rails/jobs/301276197#L2055
https://travis-ci.org/rails/rails/jobs/301276197#L2061

Because 8f2490b5 removed password from `redis-server`.
So must also remove passwords from these tests.

Add new error class `StatementTimeout` which will be raised when statement timeout exceeded (#31129)

We are sometimes using The MAX_EXECUTION_TIME hint for MySQL depending
on the situation. It will prevent catastrophic performance down by wrong
performing queries.

The new error class `StatementTimeout` will make to be easier to handle
that case.

https://dev.mysql.com/doc/refman/5.7/en/optimizer-hints.html#optimizer-hints-execution-time

Move touch method from Skipping Callbacks section to Running Callback…

[ci skip]

Verify credentials format before saving

Currently, credentials does not check the format when saving. As a result,
incorrect data as yaml is also saved.
If credentials is used in config files., an error will occur in credential
yaml parsing before edit, and will not be able to edit it.

In order to prevent this, verify the format when saving.

Related: #30851

Simplify our dev testing and CI story since we're also testing against
Redis for the Active Support cache store.

Directly test whether db, host, password, etc are passed through as
config instead of spinning up a Redis server with a password set on it.

Make automatically synchronize test schema work inside engine

Add update_only example to AR nested attributes doc [ci_skip]

It's already a default for new apps, like so many others, so
no need to flaunt it.

Allow edits of existing encrypted secrets generated on Rails 5.1,
but refer to credentials when attempting to setup.

This also removes the need for any of the setup code, so the
generator can be ripped out altogether.

Do not create credentials in dummy application

Remove unused require

Follow up of #30360.

Attribute modules (`Attribute`, `Attributes`, `AttributeSet`) uses
`Type`, but referencing `Type` before the modules still fail.

```
% ./bin/test -w test/cases/attribute_test.rb -n test_with_value_from_user_validates_the_value
Run options: -n test_with_value_from_user_validates_the_value --seed 31876

E

Error:
ActiveModel::AttributeTest#test_with_value_from_user_validates_the_value:
NameError: uninitialized constant ActiveModel::AttributeTest::Type
    /Users/kamipo/src/github.com/rails/rails/activemodel/test/cases/attribute_test.rb:233:in `block in <class:AttributeTest>'

bin/test test/cases/attribute_test.rb:232

Finished in 0.002985s, 335.0479 runs/s, 335.0479 assertions/s.
1 runs, 1 assertions, 0 failures, 1 errors, 0 skips
```

Probably we need more autoloading at least `Type`.

`joins_values.partition` will break joins values order. It should be
kept as user intended order.

Fixes #15488.

https://travis-ci.org/rails/rails/jobs/300163454#L2236

https://travis-ci.org/rails/rails/jobs/300163487#L1974