Proposed pref test fix

[andydotxyz]

The new async way of working with prefs is not OK with the test system not having complete safety of thread handling.
So take a round trip to ensure it.

Checklist:

• Tests included.
• Lint and formatter run with no errors.
• Tests all pass.

[coveralls]

coverage: 59.117% (+0.02%) from 59.094%
when pulling efd6f41 on andydotxyz:fix/preftest
into 093b53e on fyne-io:develop.

[Jacalz]

If you remove the wait group and use fyne.DoAndWait() you should get the same result. However, in this case I think we should rename the test because with this change it no longer actually tests any races but just that we get the right object (which also is a bit strange given that we add no other objects).

[andydotxyz]

(which also is a bit strange given that we add no other objects).

Because if it finds nothing another is created.
The reason this test fails is because there are lots of different instances created instead of just 1.

[andydotxyz]

If you remove the wait group and use fyne.DoAndWait() you should get the same result. However, in this case I think we should rename the test because with this change it no longer actually tests any races

Which is why I didn't do that approach - instead spawning lots of "async" fyne.Do calls as a rough approximation of what it was doing before...