Move user fields into a sub-module

[claudiu-cristea]

Fixes #81

QA notes

Clean install

Is covered by the Behat tests.

Testing the update path

Install de development environment according to the README.md.

git checkout master
./vendor/bin/run drupal:site-setup
./vendor/bin/run drupal:site-install
./vendor/bin/drush -l http://localhost/path/to/site/ uli

Go to /admin/reports/status and check that there are no Entity/Field Definitions mismatch errors.

Do a login for a test user (see the eCAS Mock Server users data) and, in the users_field_data table, check that the profile fields (field_oe_*) are correctly filled.

git checkout 81-user-fields
./vendor/bin/drush updb -y

Refresh /admin/reports/status and check that there are no Entity/Field Definitions mismatch errors.

Check the users_field_data table. The fields should be filled with the previous data.

[idimopoulos]

I tested this according to the instructions. Steps I took:

• I built a blank website according to the instructions provided in the oe_authentication module.
• I used the docker-compose.yml docker containers in order to run the authentication mock.
• In the master branch, I logged in as Chuck NORRIS, one of the mock users.
• I then logged in with an admin account and verified that there were no errors in the status report, the user Chuck NORRIS was blocked, and the base property fields retrieved and stored the values in the user account.
• Checked out the current branch and executed the updated.
• Performed the same checks to verify that everything is ok.
• Rebuilt the cache just to be sure and rechecked everything.

All tests went just like it was explained it would. I only have some commenting nitpicks.

[idimopoulos]

Adds custom...*

[idimopoulos]

We can use !== here.

[idimopoulos]

Converts an array of* EU Login....

[idimopoulos]

I changed this a bit since it was exceeding the 80 lines limit.

[idimopoulos]

Isn't it more usual in Drupal to describe this as something like "An associative array of field values indexed by the field name"?

[idimopoulos]

The exception class seems too specific for a broad exception message.

[idimopoulos]

I am removing this entirely referring back to a discussion between Antonio and Pieter about not trying to document or catch exceptions related to the entity storage operations unless we are passing them as parameters when retrieving them.

[idimopoulos]

I discussed with @claudiu-cristea and since he is already occupied with another task and these are simply nitpicks, I would implement them after him agreeing with them.

[sergepavle]

Needs rebase.