Fix test failure on PostgreSQL by sorting the result before comparison

Closes #21986.

This makes it possible to write custom types that define a different
mapping for STI columns.

Mention that halting chain does not re-raise  `ActiveRecord::RecordIn…

Make text about render :html option clearer to state, text will get e…

Mention that halting chain does not re-raise  `ActiveRecord::RecordInvalid` exception as well, similar to `ActiveRecord::Rollback`

Fixes #22297

[ci skip]

Make text about render :html option clearer to state, text will get escaped if not marked as safe by using html_safe

Fixes #22648
[ci skip]

refer to rails command instead of rake in environment error

It has been to use an overall rails command in ea4f0e2b, in order to unify.

Fix discrepancies in associations guide [ci skip]

- Followup of https://github.com/rails/rails/commit/71ff088a09d429657877ddfb58985d30df63fc8a

`ApplicationRecord` instead of  `ActiveRecord::Base` in Association G…

Forgot to remove it, when I changed the expectations in 88881d2b.

If running `bin/rails t test/models/bunny_test.rb:` we'd implicitly run all the
tests in the bunny test. I now highly doubt that people would ever put in a line
filter without a line *and* want that to mean run all tests in that file.

Instead, change regex to require a line digit after the colon, so runs without a
line at the end would fail to require the test file.

This also has the side benefit of breaking requiring a file with many colons:
`bin/rails t test/models/bunny_test.rb:::::::::::::4`

Think this means I've had enough colonoscopy to last me through the year :)

Changed options for find_each and variants to have options start/finish

CodeStatistic will now ignore hidden files:

- Modify the default regex to not match hidden files

Association Guide replaces Order as example with Author-Book Example …

fixes #21500

Add clarification about `first` and `last` behavior when using `order` [ci skip]

Checked in changes to Gemfile.lock on latest master

Updated the guides for having a distinct has_many through at the data…

[ci skip] Improve Action Cable configuration explanations

The current example

    add_index :person_articles, :article, unique: true

Does not work, the `:article` column does not exist as it will be
a join table (based on the prior example) so should use :article_id

The documentation seems to suggest that it will allow an article to
be added only once to a person via the join table, what actually
occurs is that it only allows the article to be added to one person,
at which point it should be a `belongs_to` association.

Also changed the new example to use readings based on the prior example

Explain how HStore columns can be queried [ci skip]

- Fixes #22876.

Changed options for find_each and variants to have options start/finish instead of start_at/end_at based on comments
at https://github.com/rails/rails/pull/12257#issuecomment-74688344

Pass through AP CHANGELOG

* Clarify wording around how defaults work
* Clarify wording around what formatting versus information is required
* Separate consumer configuration section since it is important and wordy
* Add additional explanation of options for setting consumer configuration
* Consolidate consumer configuration information to reduce confusion

Remove AV dependence on AP's Mime

Enrich Action View Template types with symbols from Action Dispatch Mime.

A Set can't be implicitly converted into an Array:

```
irb(main):012:0> formats = [ :rss ]
=> [:rss]
irb(main):013:0> formats &= SET.symbols
TypeError: no implicit conversion of Set into Array
    from (irb):13:in `&'
    from (irb):13
    from /Users/kasperhansen/.rbenv/versions/2.2.3/bin/irb:11:in `<main>'
```

Besides `Mime::SET.symbols` returns an Array, so we're closer to that.

Match `Mime::SET.symbols`.

If they aren't symbols, then they aren't likely to be in the set anyway.

We'll be using this to map over to Action Dispatch's Mime::Set.

The template types is a private abstraction to fill in basic blanks from Action Dispatch's
mime types. As such we can modify the data structure ourselves.

Spares a to_sym call by aliasing to_sym to ref.

Then the delegate felt meager for one method; ditch and define method ourselves.