fix typo

Fixes #23645

When you're using an `attr_accessor` for a record instead of an
attribute in the database there's no way for the record to know if it
has `changed?` unless you tell it `attribute_will_change!("attribute")`.

The change made in 27aa4dda updated validations to check if a record was
`changed?` or `marked_for_destruction?` or not `persisted?`. It did not
take into account virtual attributes that do not affect the model's
dirty status.

The only way to fix this is to always validate the record if the
attribute does not belong to the set of attributes the record expects
(in `record.attributes`) because virtual attributes will not be in that
hash.

I think we should consider deprecating this particular behavior in the
future and requiring that the user mark the record dirty by noting that
the virtual attribute will change. Unfortunately this isn't easy because
we have no way of knowing that you did the "right thing" in your
application by marking it dirty and will get the deprecation warning
even if you are doing the correct thing.

For now this restores expected behavior when using a virtual attribute
by always validating the record, as well as adds tests for this case.

I was going to add the `!record.attributes.include?(attribute)` to the
`should_validate?` method but `uniqueness` cannot validate a virtual
attribute with nothing to hold on to the attribute. Because of this
`should_validate?` was about to become a very messy method so I decided
to split them up so we can handle it specifically for each case.

Fixed a typo

habtm join tables commonly have two id columns and it's OK to make those
two id columns a primary key.  This commit eliminates the warnings for
join tables that have this setup.

  ManageIQ/manageiq#6713

Deprecate AC::Parameters#== with Hash

Add methods back to array delegation from ActiveRecord::Relation

Fix button_to's params option to support nested names.

Delegation of some `Array` methods was removed in commit 9d79334a. That
change did add explicit delegation of a few methods that `Array` has but
which aren't on `Enumerable`.  However, a few non-mutation methods were
omitted.  This adds `Array` delegation of `#in_groups`, `#in_groups_of`,
`#shuffle` and `#split`.  This allows things like
`MyThing.all.in_groups_of(3) { ... }` to continue working as they did
before commit 9d79334a.

Dependencies clean up

In e6e0579d the `params` option was added to the `button_to` helper. However, the patch doesn't support nested hashes so `{a: {b: 'c'}}` for example gets turned into a hidden form input with the name 'a' and the value being the string representation of the `{b: 'c'}` nested hash.

Since Rails supports nested hashes everywhere else (and even in the URL params of link_to and button_to), I believe this to be a bug/unfinished feature.

Make ActionCable logging less verbose in development

Cleans up four items I came across in ActiveSupport::Dependencies:

- DependenciesTest#
  test_dependency_which_raises_exception_isnt_added_to_loaded_set:
  Fixes current implementation which will pass no matter what since the
  filepath is never added to "loaded" or "history" without being
  expanded first.
- Remove DependenciesTest#test_unhook. Seems leftover from when
  alias_method_chain was used in Loadable and ModuleConstMissing.
  The test will always pass since Module never responds to those methods
- WatchStack#new_constants documentation: update self to @stack.
  Looks like self was leftover from when WatchStack inherited from Hash
- Remove ActiveSupport namespace from call to
  Dependencies.constant_watch_stack.watching? since the namespace is not
  needed, Dependencies is called two other times in the same method
  without it (even on the same line) and it brings the line to within
  80 characters

Creating a protected getter method for `@parameters`.

Merge pull request #23773 from wisetara/wisetara/update-docs-ActiveSupport__TestCase#assert_nothing_raised-for-pr

better docs for ActiveSupport::TestCase#assert_nothing_raised

remove needless `debug_exception_response_format` config [ci skip]

Since a0343d11, `debug_exception_response_format` config depends on `api_only`.
Therefore, if set the `api_only`, need to specify `debug_exception_response_format` is not.

Remove unused require

Fix `unsubscribed` server side behavior

Before this commit, the `unsubscribed` callbacks in Action Cable server
side channels were never called. This is because when a WebSocket
"goodbye" message was sent from the client, the Action Cable server
didn't properly clean up after the now closed WebSocket. This means that
memory could possibly skyrocket with this behavior, since part of this
commit is to properly remove closed subscriptions from the global
subscriptions hash. Say you have 10,000 users currently connected, and
then all 10,000 disconnect -- before this patch, Action Cable would
still hold onto information (and Ruby objects!) for all of these now
dead connections.

The Event hack was too much of a hack: on actually thinking about it,
there's a rather obvious race.

Fix master build

All gems are released now so we don't need to ask the github repository
for these gems anymore.

move digest cache on to the DetailsKey object

set `skip_listen` option to dummy appplication

This moves digest calculation cache on to the details key object.
Before, the digest cache was a class level ivar, and one of the keys was
the hash value of the details key object:

  https://github.com/rails/rails/blob/13c4cc3b5aea02716b7459c0da641438077f5236/actionview/lib/action_view/digestor.rb#L28

An object's hash value is not unique, so it's possible for this cache
key to produce colliding keys with no resolution.  This commit move
cache on to the details key object itself, so we know that the digests
are always unique per details key object.

I don't think caching this method makes any difference on Ruby 2.0:

```
require 'benchmark/ips'

class Foo
  alias :object_hash :hash
  attr_reader :hash
  def initialize
    @hash = object_hash
  end
end

class Bar
end

hash = {}
foo = Foo.new
bar = Bar.new
Benchmark.ips do |x|
  x.report("foo") { hash[foo] }
  x.report("bar") { hash[bar] }

  x.report("foo.hash") { foo.hash }
  x.report("bar.hash") { bar.hash }
end

__END__
[aaron@TC ruby (trunk)]$ ruby test.rb
Warming up --------------------------------------
                 foo   118.361k i/100ms
                 bar   118.637k i/100ms
Calculating -------------------------------------
                 foo      7.944M (± 3.1%) i/s -     39.769M
                 bar      7.931M (± 3.4%) i/s -     39.625M
[aaron@TC ruby (trunk)]$ ruby test.rb
Warming up --------------------------------------
                 foo   122.180k i/100ms
                 bar   120.492k i/100ms
            foo.hash   123.397k i/100ms
            bar.hash   119.312k i/100ms
Calculating -------------------------------------
                 foo      8.002M (± 4.2%) i/s -     39.953M
                 bar      8.037M (± 4.5%) i/s -     40.124M
            foo.hash      8.819M (± 3.9%) i/s -     44.053M
            bar.hash      7.856M (± 4.1%) i/s -     39.254M
```

fix the language for simple and proper understanding

When running the ActionCable server in development I get a lot of output
in my logs, this commit sets a maximum length of 300 characters for a
broadcast log message.

Missing documentation about hash algorithm option for MessageVerifier

If you want to use `EventedFileUpdateChecker`, need to specify `listen` to plugin's gemspec.
However, the default is not specified `listen` to plugin's gemspec,
`EventedFileUpdateChecker` should be disabled.