Log active_job potential matches when asserting

[garethson]

Adds logging of potential matches when calling assert_enqueued_with and assert_performed_with to provide more information on test failures.

Summary

When testing ActiveJob, it's quite helpful to view the jobs that were performed (or enqueued) when these assertions fail.

Before:

# Running:

...........F

Failure:
PerformedJobsTest#test_assert_performed_with_selective_args [/Users/garethduplooy/src/github.com/rails/activejob/test/cases/test_helper_test.rb:1761]:
No performed job found with {:job=>MultipleKwargsJob, :args=>{:argument2=>{:b=>3, :a=>4}, :argument1=>1}}

After:

# Running:

...........F

Failure:
PerformedJobsTest#test_assert_performed_with_selective_args [/Users/garethduplooy/src/github.com/rails/activejob/test/cases/test_helper_test.rb:1761]:
No performed job found with {:job=>MultipleKwargsJob, :args=>{:argument2=>{:b=>3, :a=>4}, :argument1=>1}}
Potential matches: {:job=>MultipleKwargsJob, :args=>[{:argument2=>{:b=>2, :a=>1}, :argument1=>1}], :queue=>"default"}

The added context makes it much easier to determine the source of failure (wrong job, jobs, wrong params, wrong queue, etc).

cc @gmcgibbon @rafaelfranca

[gmcgibbon]

I think << is cheaper right? I would also prefer if potential_matches.any? here.

[itsWill]

note since the # frozen_string_literal: true comment is enabled, << will fail, in which case you can do:

message = +"No enqueued job found with #{expected}"

And then << should work correctly.

[gmcgibbon]

I think I would also prefer a line break in the output:

# Running:

...........F

Failure:
PerformedJobsTest#test_assert_performed_with_selective_args [/Users/garethduplooy/src/github.com/rails/activejob/test/cases/test_helper_test.rb:1761]:
No performed job found with {:job=>MultipleKwargsJob, :args=>{:argument2=>{:b=>3, :a=>4}, :argument1=>1}}

Potential matches: {:job=>MultipleKwargsJob, :args=>[{:argument2=>{:b=>2, :a=>1}, :argument1=>1}], :queue=>"default"}