Pass default_url_options from mailer

[djfpaagman]

This is a bit related to #187, with the same behavior where we set the hostname dynamically based on locale.

With the introduction of the context in 1.1.0 (I think?) (PR: #180). the url helpers are no longer using the default_url_options from the mailer (either via config.action_mailer.default_url_options or when overriding them in a mailer themselves via a default_url_options method).

This PR fixes that by passing them as arguments to the url_for method from the context.

This probably also means you no longer necessarily need to to set a default host for the routes as well, as mentioned here in the docs:

passwordless/README.md

Lines 148 to 155 in cd0e0ed

	Also be sure to
	[specify ActionMailer's `default_url_options.host`](http://guides.rubyonrails.org/action_mailer_basics.html#generating-urls-in-action-mailer-views) and tell the routes as well:
	```ruby
	# config/application.rb for example:
	config.action_mailer.default_url_options = {host: "www.example.com"}
	routes.default_url_options[:host] ||= "www.example.com"
	```

I also added a reload when using WithConfig, as the parent class for the mailer is set from the config and that can change based on usage of the helper. This probably also needs to be included in #187.

Let me know what you think!

[mikker]

Great solution!

[mikker]

The tests are failing. I think maybe due to your refresh mechanism not working as intended?

[djfpaagman]

I had some troubles running the tests locally (fixed that now) and indeed this does not seem to work without side effects. I'll have another look at it at a later point!

[djfpaagman]

I think I found a different solution, tests are constantly green for me now!

[mikker]

At least it's green! I think it'd be better to move all this from the general with_config to the specific test? It is only really relevant to these two examples.

[djfpaagman]

Yeah that could work as well 👍 but I think there will always be the possibility that a future test needs to override parent_mailer as well and then you'll run into the same issue again. So I think keeping the reloading behavior in the test helper for overriding configs makes most sense to me.

But happy to move it to the test itself as well, let me know.

[mikker]

Let's move it now and move it back when it becomes necessary. Thanks!

[djfpaagman]

@mikker: I've moved some things around with comments so that it's clear what's going on.

[djfpaagman]

@mikker is there anything I can do to get this merged in? would love to see it released :)

[mikker]

Sorry, no, it's good. Thank you!