On a local install and on a test server or in headless mode the default
window size may change which may cause different behaviours.

We want to see errors.

I want to see what happens when make functional_browser succeeds.

We should wait.

The title of the return window has changed.
Moreover, there was a 'rescue' in the test, this was a bad idea, hiding the actual cause of the error.

$b.unselect was not working, we just comment it and do a workaround in some cases

Now the number of reads may have spaces (243 241)

This happens when the server address is wrong.

The query name depends on the results of the previous tests.
The length should be always 119.

This is the first test that will fail if a bad index.html is given.

Some webdrivers require such canonical paths.

Class VidjilBrowser knows how to get the elements in the browser (it knows the class, the hierarchy and so on).
Inside the tests we should rely as much as possible on VidjilBrowser rather than using directly class names, ID or texts
to retrieve a given element.

Of course VidjilBrowser should be completed all along the time :)

The same should happen when hovering a clone in the list or in the scatterplot (also for clicking).
Therefore we rely on the same function to test what should happen.

Remove tests that were using the visibility of the graph

The text may change not the ID

Now the headless server is launched just once as well as the browser.
The browser can be setup again if needed (to reset everything for instance).

The test_zz is the test that must be launched as last test to close browser and Xvfb

The text may change not the ID

Now the headless server is launched just once as well as the browser.
The browser can be setup again if needed (to reset everything for instance).

The test_zz is the test that must be launched as last test to close browser and Xvfb

Data file used is the first one from the doc (automatically retrieved from org file).
The browser test now uses a headless browser if HEADLESS environment variable is defined.

- remove custom runner (deprecated) use miniTest::spec and after_run instead
- instal gem minitest 5.3.1 and minitest-ci 3.0.3 to use it
- generate Xunit report