Without this, bundler tries to bundle json 1.8 on Ruby 2.4 in some cases
https://travis-ci.org/rails/rails/jobs/175874852

Ruby Kindle periodical-format ebook generator

When the initial evented monitor feature was written, the latest version of listen
was the 3.0.x series. Since then the listen project has moved on to the 3.1.x series.
This patch allows the use of the new versions.

Mostly, this is just to avoid EventMachine. But there's also an argument
to be made that we're better off using a different protocol library for
our test suite than the one we use to implement the server.

- This is required for bundler 1.13.0 because of which lot of specs are
  failing on Travis CI.
- Similar to https://travis-ci.org/rails/rails/jobs/158905576#L559

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

See https://github.com/nex3/rb-inotify/pull/49

When run test of Active Job with resque 1.26, occurs following error.

```
QueuingTest#test_current_locale_is_kept_while_running_perform_later:
NoMethodError: undefined method `current_tags' for #<Resque::QuietFormatter:0x0055b44f63ed50>
    /home/yaginuma/program/rails/master_y_yagi/rails/activejob/lib/active_job/logging.rb:51:in `logger_tagged_by_active_job?'
```

This was happening for the formatter class of resque not the formatter class of
Rails is they've been used to logger.formatter.
This was happening because become fomatter is changed during the instantiation
of worker in the resque 1.26.
In the master, unless the environment variable is set, fomatter is so as not to
be changed, test will pass.

Ref: https://github.com/resque/resque/pull/1439

per security release today --> https://groups.google.com/forum/#!topic/ruby-security-ann/RCHyF5K9Lbc

Allow failures until test runs are consistently stable, not hanging.

Closes #24943.
Signed-off-by: NJeremy Daer <jeremydaer@gmail.com>

Apps that depend on Action Cable don't need Blade for app development,
so we can remove the gem dependency.

We do need Blade for Action Cable dev, so we bundle it in the Gemfile.
Signed-off-by: NJeremy Daer <jeremydaer@gmail.com>

Without the `wdm` gem, it appears that `listen` keeps an open handle to
each of these files, causing them not to be removed when the tempdir
tries to clean iteslf up, and then directory to fail to unlink. In
addition to fixing that particular failure, we now construct OS agnostic
paths, and capture exceptions if the directory fails to unlink so that
minitest will report it rather than crash

Uses latest precompiled JRuby so that
we don't spend time downloading versions Travis has
not already compiled. http://rubies.travis-ci.org/

Uses latest jdk: oraclejdk8
per
https://docs.travis-ci.com/user/build-environment-updates/2015-02-03/#Ruby-VM
and
https://docs.travis-ci.com/user/languages/ruby/#Supported-Ruby-Versions-and-RVM

Follows on work in https://github.com/rails/rails/pull/23927 which was reverted
https://github.com/rails/rails/commit/26fe5fa08d52384bcea09c4210e3b3baaf5e9b95

JRUBY_OPTS minimize GC, disable JIT, for max test speed
  - https://github.com/rails/rails/pull/16613
  - https://github.com/rails/rails/pull/17088

Have Rails use JRuby-compatible Rake 11.1
  - The Rake task was passing --verbose, an invalid option, to contemporary JRuby
  - https://github.com/ruby/rake/pull/120
  - https://github.com/rails-api/active_model_serializers/pull/1585
  - https://github.com/jruby/jruby/issues/3653#issuecomment-195883717

No advantage to directly mounting JRuby over installing from cache; both on S3
  - https://github.com/rails/rails/pull/23499/commits/b2d5b336b5a812009444571b22eed90777cd9a8c
  - https://github.com/rails/rails/pull/23499/commits/f4fad041b2755d81e82ee3fc66d96f334b4653db

This undoes https://github.com/rails/rails/commit/7241498e51120b9847a8bc16cf48551db0f3e216

https://github.com/codahale/bcrypt-ruby/issues/128 is fixed and closed.

Changes "as has to be loaded" to "as it has to be loaded"

Ref - https://github.com/turbolinks/turbolinks-rails/pull/3

All modern Rubies ship JSON as part of stdlib.  Using the gem actually hurts multi-platform support due to build difficulties on Windows.

This new adapter does get a little more intimate with the redis-rb gem's
implementation than I would like, but it's the least bad of the
approaches I've come up with.