Bugfix: unscope(where: [columns]) would not remove the correct binds

Catch postgres connection errors when trying to dealloc

Add `except` option for ActiveJob::TestHelper methods

Fix unscoping `default_scope` in STI associations

Dates are able to be natively handled by the mysql2 gem. libmysql (and
the wire protocol) represent each portion of the date as an integer,
which is significantly faster to encode and decode. By passing the Ruby
date objects through directly, we can save a good bit of time and
memory.

Since 5c71000d, it has lost to be able to unscope `default_scope` in STI
associations. This change will use `.empty_scope?` instead of
`.values.empty?` to regard as an empty scope if only have
`type_condition`.

Fix typo s/rause/raise/

This method needs to be protected.

Fix `JoinDependency` with using a custom table

[ci skip] CodeTriage badge

[Code Triage](https://www.codetriage.com/) is an app I've maintained for the past 4-5 years with the intent of getting people involved in open source. It sends subscribers a random open issue for them to help "triage". For Ruby projects such as "rails/rails" you can also subscribe to documentation. For example you can get a few random documented methods, or if you want to write docs, get undocumented methods.

The initial approach was inspired by seeing the work of the small core team spending countless hours asking "what rails version was this in" and "can you give us an example app". The idea is to outsource these small interactions to a huge team of volunteers and let the core team focus on their work.

The purpose of the badge is to give more people an easier way to start contributing to Rails. Here's what it currently looks like:

[![Code Triage Badge](https://www.codetriage.com/rails/rails/badges/users.svg)](https://www.codetriage.com/rails/rails)

The number is how many people are currently subscribed (a.k.a. "helpers") to the project on CodeTriage, the color is based off of the number of open issues in the project. You can see an example of this badge on another popular open source repo [Crystal](github.com/crystal-lang/crystal/).

> For context to non-rails core: I also maintain sprockets (though a release hasn't happened in some time, sorry), and I have commit to Rails. I'm not some rando trying to push arbitrary links to READMEs on GitHub.

support `-` as an argument to `rails runner`

in Rails 4.0, you could use `/dev/stdin` on both Linux and Mac, but with
the switch to Kernel.load in Rails 4.1, this broke on Linux (you get
a LoadError). Instead, explicitly detect `-` as meaning stdin, then
read from stdin explicitly, instead of performing file gymnastics. This
should now work on any platform uniformly.

Passing a script via stdin is useful when you're sshing to a server,
and the script you want to run is stored locally. You could theoretically
pass the entire script on the command line, but in reality you'll run
into problems with the command being too long.

Clarifies the intent that aren't just loading the
model but really caring about triggering the on_load
callbacks.

Without this fix, `JoinDependency` doesn't use a custom table alias:

```
% ARCONN=sqlite3 be ruby -w -Itest test/cases/relations_test.rb -n test_using_a_custom_table_with_joins_affects_the_wheres
Using sqlite3
Run options: -n test_using_a_custom_table_with_joins_affects_the_wheres --seed 14531

E

Error:RelationTest#test_using_a_custom_table_with_joins_affects_the_wheres:
ActiveRecord::StatementInvalid: SQLite3::SQLException: no such column: posts.author_id: SELECT  "omg_posts".* FROM "posts" "omg_posts" INNER JOIN "authors" ON "authors"."id" = "posts"."author_id" WHERE "omg_posts"."title" = ? LIMIT ?
```

Extract `NullRelationTest` from `RelationTest`

Remove outdated `test_scoped_responds_to_delegated_methods`

Correctly handle frozen options for ActiveRecord::Serialization#seria…

Fix HashWithIndifferentAccess#default when include?(nil)

Fix `where` with a custom table

`Persistence#delete` should not be affected by scoping

The implementation of HashWithIndifferentAccess#default didn't
distinguish `default` from `default(nil)`, which caused an incorrect
result for `default` if `nil` was used as a key.

Define HashWithIndifferentAccess#dig so that hackery that behaves
differently from Hash#default can be removed from
HashWithIndifferentAccess#default.

Without this fix, SELECT clause doesn't use a custom table alias name:

```
% ARCONN=sqlite3 be ruby -w -Itest test/cases/relations_test.rb -n test_using_a_custom_table_affects_the_wheres
Using sqlite3
Run options: -n test_using_a_custom_table_affects_the_wheres --seed 31818

E

Error:
RelationTest#test_using_a_custom_table_affects_the_wheres:
ActiveRecord::StatementInvalid: SQLite3::SQLException: no such table: posts: SELECT  "posts".* FROM "posts" "omg_posts" WHERE "omg_posts"."title" = ? LIMIT ?
```

`self.class.delete` is delegated to `all` and `all` is affected by
scoping. It should use `unscoped` to not be affected by that.

Add bootsnap to default Gemfile

Remove unused `Mutex_m` in Active Model

Scrub the invalid paramter value before using it in the error

Bootsnap precomputes load path resolution and caches ruby ISeq
and YAML parsing/compilation, reducing application boot time by
approximately 50% on supported configurations.

You should be able to safely use the String error message. So when
finding the paramter has an invalid encoding we need to remove the
invalid bytes before using it in the error. Otherwise the caller might
get another Encoding error if they use the message.

Fixing  "Adding Member Routes" documentation

Between 4.2 and 5.0 the behavior of how multiparameter attributes
interact with `_before_type_cast` changed. In 4.2 it returns the
post-type-cast value. After 5.0, it returns the hash that gets sent to
the type. This behavior is correct, but will cause an issue if you then
tried to render that value in an input like `text_field` or
`hidden_field`.

In this case, we want those fields to use the post-type-cast form,
instead of the `_before_type_cast` (the main reason it uses
`_before_type_cast` at all is to avoid losing data when casting a
non-numeric string to integer).

I've opted to modify `came_from_user?` rather than introduce a new
method for this as I want to avoid complicating that contract further,
and technically the multiparameter hash didn't come from assignment, it
was constructed internally by AR.

Close #27888.

Add documentation for class_attribute default option

Remove useless `arel_engine`