Don't use remove_method or remove_possible_method just before a new
definition: at best the purpose is unclear, and at worst it creates a
race condition.

Instead, prefer redefine_method when practical, and
silence_redefinition_of_method otherwise.

This reverts commit 3420a145, reversing
changes made to afb66a5a.

* Allow a default value to be declared for class_attribute

* Convert to using class_attribute default rather than explicit setter

* Removed instance_accessor option by mistake

* False is a valid default value

* Documentation

The class-level respond_to was extracted in ee77770d to responders gem

[ci skip]

The current code base is not uniform. After some discussion,
we have chosen to go with double quotes by default.

[ci skip]

To complement actionpack/test/controller/metal/renderers_test.rb

Rails 4.x and earlier didn't support `Mime::Type[:FOO]`, so libraries
that support multiple Rails versions would've had to feature-detect
whether to use `Mime::Type[:FOO]` or `Mime::FOO`.

`Mime[:foo]` has been around for ages to look up registered MIME types
by symbol / extension, though, so libraries and plugins can safely
switch to that without breaking backward- or forward-compatibility.

Note: `Mime::ALL` isn't a real MIME type and isn't registered for lookup
by type or extension, so it's not available as `Mime[:all]`. We use it
internally as a wildcard for `respond_to` negotiation. If you use this
internal constant, continue to reference it with `Mime::ALL`.

Ref. efc6dd55

We should be asking the mime type method for the mime objects rather
than via const lookup

of respond_to. respond_with was moved into the responders gem and deprecated
inside rails, so there is no need to mention it within rails itself.

* Renames _handle_render_options to _render_to_body_with_renderer, which is more
  intention-revealing.

* The name of the dynamically generated method for a renderer with key :js was
  "_render_option_js".

  That name is too weak. :js is an option if you see the render argument as just
  a generic options hash, but in the context of renderers that's the renderer
  key, is what identifies the renderer.

  Now "_render_with_renderer_js" is generated instead, which is crystal clear.

* The name of the dynamically generated method for the renderer was constructed
  using string literals in a few places. That is now encapsulated in a method.

* Since we were on it, also removed a couple of redundant selfs.

Adds a comment before JSONP callbacks. See
http://miki.it/blog/2014/7/8/abusing-jsonp-with-rosetta-flash/ for more
details on the exploit in question.

The controller can set the response format as 'JSON' before the renderer code be
evaluated, so we must replace it when necessary.

Fixes #15081

ActionController::Renderers::RENDERERS is an instance of Set. Docs incorrectly
state that it's a Hash.

* Add "<tt>" or "+" to improve font of some code and filenames in API documentation
* Does not contain wording changes

Currently if a user calls #respond_with(csvable), but has not csv
renderer available, Responder will just run through the default render
behavior twice, raising ActionView::MissingTemplate both times.

This changes ActionController::Metal::Responder#api_behavior to check
in advance whether there is a renderer available, and raise
ActionController::MissingRenderer if not.

If a user wants json output then try best to render json output. In such cases prefer kind_of(String) over respond_to?(to_str)

[#5841 state:resolved]
Signed-off-by: NJosé Valim <jose.valim@gmail.com>