According to the requests of the Engineering Diffraction scientists.

If custom spectra was selected from the dropdown and then the tick box
was deselected, the GUI would still check the validity of the spectrum
numbers. As this was usually empty, the calibration would fail to run.

Also update presenter to fix minor bugs found while testing.

The functionality was the same across calibration and focus.

Currently, there is no way to load a CSV file (which is the format
that the full calibration is in) into mantid to be used by the GUI.
Until that is added, the option to select the full calibration has been
hidden from the user.
The functionality is there, so once the loading works for large CSVs the
hide() call can be removed.

- Remove repeated searching checks

- Split test into reasonable chunks

- Make rb_num naming consistent

Vanadium output files are now saved to the user directory if an RB
number is specifies so that users can have all of the generated data
from processing.

Focusing now saves as .gss, .nxs, and .dat files.

Now using sample instead of a mix of both of these.

Before, the file finders (MWRunFiles) could only be made read-only,
which did not grey out the text field, just the browse button. This
option has been added and used instead for the file inputs that should
not be used.

Loading the vanadium to be used later on makes more sense to do at this
stage than at the start of the focusing step.

The calibration parameter table that is generated when creating a new
calibration is now loaded back into mantid from the file as a table
workspace, rather then just working in the background.

Parts of the UI that do not do anything are now greyed out to avoid
confusion. The returned values from the load are also added to the new
calibration boxes to show that the load has succeeded.

Loading an existing calibration from the GSAS parameter files now reloads the calibration from the information stored in the file.

Also refactors the tests to focus on one path of execution per test.

Debug error appeared on windows when buttons are enabled at the end of
the calibration and focus.

By emitting a signal instead of calling the function to enable the
buttons directly, windows stops complaining about threads interacting
with the GUI.

Documentation has been added for the less obvious public and private methods.
Some of the methods in the model should be shown as private, so have
had methods refactored.

Test file names need to be distinct, as CMake does not
 allow for build properly when file names are the
 same.  Also adds main methods to test classes.