chore(test): clean integration test suite

[jniles]

This commit cleans up the application's integration test suite in
several ways:

1. Ensure all tests using UUIDs are using node-uuid, and ensuring
   they are using it properly (uuid.v4() over uuid()).
2. Removes extraneous comments.
3. Transitions many modules to using test helpers exclusively. Removes
   the errorKeys property from the test helpers -- all errors should be
   caught via helpers.api.errored().
4. [perf] Improves testing logs and performance by using before() instead of beforeEach() to log the agent in.

The goal is to have cleaner, more maintainable tests for developers to
use as references when writing future integration tests.

[DedrickEnc]

I am happy to notice that, you deleted used test files

The integration test was difficult at the beginning because every one was supposed to learn the API, but now every one is so quick, and the same situation is observable for the end to end test, it is so difficult but every one will be quick with the time.

You introduces new interfaces to the helpers.js file.

So I want to know, since this change will affect the productivity of every one :

1. What is the motivation which leads you to almost rewrite the integration test?
2. Did you discuss that with the team? (did you at least create an issue?)
3. In some situation testing an API can necessitate some particulars checks which are very tailed to the business logic, do you think these generics methods (listed, errored ...) are enough to satisfy every developers during the test?

e.g : user.js
without helpers.api

expect(res).to.have.status(200);
expect(res.body).to.not.be.empty;
expect(res.body).to.have.length(1);     
expect(res.body[0]).to.have.keys('id', 'unit_id');
expect(res.body[0].unit_id).to.equal(0);

with helpers api

helpers.api.listed(res, 1);
expect(res.body[0]).to.have.keys('id', 'unit_id');
expect(res.body[0].unit_id).to.equal(0);

4.. Why some files are not entirely covered by the change, e.g employee.js

[jniles]

Motivation
Everyone implemented a slightly different method of testing. Some developers checked for json, others did not. Some developers checkout .to.have.length(N), others simply check .not.empty. There was was a lot of boilerplate code.

The helper methods were developed to allow our team to focus on testing situations, rather than simply writing lots of code. Since most APIs behave the same (all return JSON encoding, create statements should send back either an ID or a UUID, etc...), it makes sense to move these checks into a place they can be used easily and frequently.

No, I did not discuss it with the team. That discussion is free to happen here. There are many times when I've created issues that are never read by @IMA-WorldHealth/local-contributors , so I am experimenting with new ways to show my ideas.

Of course generic methods will not be enough to replace specific cases. Our development team should still be thinking about, and writing specific tests. This just helps write common tests faster.

Some files are not entirely covered because I didn't get around to completing them.

It is evident that these types of changes are not appreciated. So, I am closing this pull request.