Fix optimized url helpers when using relative url root

Fixes #31220.

[ci skip] Update MVC wiki link

Follow up of 924a368f

Include migration files in gem

Fixes #31245

There are two concerns which are both being combined into one here, but
both have the same goal. There are certain attributes which we want to
always consider initialized. Previously, they were handled separately.
The primary key (which is assumed to be backed by a database column)
needs to be initialized, because there is a ton of code in Active Record
that assumes `foo.id` will never raise. Additionally, we want attributes
which aren't backed by a database column to always be initialized, since
we would never receive a database value for them.

Ultimately these two concerns can be combined into one. The old
implementation hid a lot of inherent complexity, and is hard to optimize
from the outside. We can simplify things significantly by just passing
in a hash.

This has slightly different semantics from the old behavior, in that
`Foo.select(:bar).first.id` will return the default value for the
primary key, rather than `nil` unconditionally -- however, the default
value is always `nil` in practice.

Since #31173, mysql2 adapter depends on `automatic_close` which is
introduced since mysql2 0.4.3. So the adapter with the mysql2 version
before doesn't work with fork now.

```
% ARCONN=mysql2 be ruby -w -Itest test/cases/connection_adapters/connection_handler_test.rb -n test_forked_child_doesnt_mangle_parent_connection
Using mysql2
Run options: -n test_forked_child_doesnt_mangle_parent_connection --seed 19988

/Users/kamipo/src/github.com/rails/rails/activerecord/lib/active_record/connection_adapters/mysql2_adapter.rb:108:in `discard!': undefined method `automatic_close=' for #<Mysql2::Client:0x00007fedaa91dfd0> (NoMethodError)
```

This drops mysql2 version less than 0.4.3 to guarantee fork safety.

[ci skip] Fix documentation for deprecation method_wrappers

Update Rails on Rack guide [ci skip]

Seems like version with class methods doesn't work. 

```ruby
require "active_support/deprecation.rb"

module Fred
  extend self

  def aaa; end
  def bbb; end
  def ccc; end
  def ddd; end
  def eee; end
end

ActiveSupport::Deprecation.deprecate_methods(Fred, :aaa, bbb: :zzz, ccc: 'use Bar#ccc instead')
Fred.aaa
Fred.bbb
Fred.ccc
```
# produces nothing

vs

```ruby
require "active_support/deprecation.rb"

class Fred
  def aaa; end
  def bbb; end
  def ccc; end
  def ddd; end
  def eee; end
end

ActiveSupport::Deprecation.deprecate_methods(Fred, :aaa, bbb: :zzz, ccc: 'use Bar#ccc instead')
Fred.new.aaa
Fred.new.bbb
Fred.new.ccc
```

produces

```
DEPRECATION WARNING: aaa is deprecated and will be removed from Rails 5.2 (called from <main> at deprications.rb:15)
DEPRECATION WARNING: bbb is deprecated and will be removed from Rails 5.2 (use zzz instead) (called from <main> at deprications.rb:16)
DEPRECATION WARNING: ccc is deprecated and will be removed from Rails 5.2 (use Bar#ccc instead) (called from <main> at deprications.rb:17)
```

Added first draft of Rails 5.2 release notes [ci skip]

This will keep the behavior of an application with the defaults of a 4.2
or 5.0 application behaving the same when upgrading to 5.2.

Since `ContentSecurityPolicy::Middleware` is not loaded in API-only
Applications, initializer is unnecessary.
Ref: https://github.com/rails/rails/blob/9c10fec4c06da38f8975dfb851f4d899aa85f8b7/railties/lib/rails/application/default_middleware_stack.rb#L66..L68

Added missing test for presence of CSP initializer

Fix CHANGELOG for CSP PR #31162 [ci skip]

- Followup of #31162

Add DSL for configuring Content-Security-Policy header

The def with blank `()` was newly added in #31176, but we have not used
the blank `()` style in most part of our code base.
So I've enabled `Style/DefWithParentheses` to prevent to newly added the
code.

https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Content-Security-Policy

Use Puma 3.11 in newly generated applications

In order to use early hints, need to use Puma 3.11.0 or higher.
So, I think that should specify that version in newly generated applications.

Ref: https://github.com/puma/puma/commit/f6f3892f4d82638fb7a2a57d993641b1486ee88a

Clean database naming ambiguity

Add new error class `QueryCanceled` which will be raised when canceling statement due to user request (#31235)

This changes `StatementTimeout` to `QueryCanceled` for PostgreSQL.

In MySQL, errno 1317 (`ER_QUERY_INTERRUPTED`) is only used when the
query is manually cancelled.

But in PostgreSQL, `QUERY_CANCELED` error code (57014) which is used
`StatementTimeout` is also used when the both case. And, we can not tell
which reason happened.

So I decided to introduce new error class `QueryCanceled` closer to the
error code name.

Since #31129, new error class `StatementTimeout` has been added.
`TransactionTimeout` is caused by the timeout shorter than
`StatementTimeout`, but its name is too generic. I think that it should
be a name that understands the difference with `StatementTimeout`.

Remove field ids from scaffold form

This reverts commit 0f79ab91, reversing
changes made to d575f7f2.

This PR philosophically conflicts with #30800 and Matthew thinks we
should hold off merging this until we find concensus. Reverting since
we're about to cut a release for 5.2.

In some places this file referred to the database in three different ways: database, DB and db. The last one caused confusion with the db namespace and the db folder. This commit removes this ambiguity by using the whole word 'database' everywhere

This reverts commit 536d3068.

Flush idle database connections