Why is email uniqueness validation performed in such a strange way?

1. I agree with #613 issue - overriding `email_changed?` is risky because this method is a part of `ActiveModel::Dirty`, which could be quite useful.
2. Why do you write your own `unique_email_user` validator for email validation instead of using `ActiveRecord` built-in `UniquenessValidator`? It makes quite inconvenient to, for example, implement uniqueness validation also on update, not only create like it is now. I had to override `unique_email_user` validation method to achieve this:

``` ruby
# here's my implementation of uniqueness validation, working also on update:
validates_uniqueness_of :email, scope: :provider, if: -> { provider == 'email' }

# ...

# and here's the `unique_email_user` method I had to override:

def unique_email_user
true
end

```

Start by reading the unique_email_user validation and the email_changed? behavior referenced in the issue, along with ActiveRecord's UniquenessValidator and issue #613. The work is complete when the project has a decided validation approach, with create and update uniqueness behavior covered by tests.

Written by the indexing model from the issue text.