use dialog fixtures and fix flaky tests

[sssoleileraaa]

Description

Fixes #1139

Test Plan

Confirm #1139 is fixed.

• Check CI: https://app.circleci.com/pipelines/github/freedomofpress/securedrop-client?branch=1139-fix
  • confirm there are 10+ successful runs in a row for this PR branch (when CI used to fail ~50% of the time)
• run script from STR section of Investigate flaky tests #1139 (takes approx 17 minutes if there are no failures: 90-120 seconds each test run)
  • confirm no dialog errors
• modify script to execute make test instead of make test-random and modify how many times you loop if you're not satified with 10 (I've tested with 100)
• run script (takes approx 7 minutes for 10 loops if there are no failures: 30-50 seconds each test run)
  • confirm no dialog errors

[sssoleileraaa]

while super focused on flaky test errors I forgot about linter errors. ok, now this is ready for review.

[emkll]

Thanks @creviera for opening this, the tests do appear to be more reliable. With make test, tests are reliably passing. However, getting intermittent failures with make test-random:

________________________________________________________________________________________________________ test_SourceList_set_snippet _________________________________________________________________________________________________________
CALL ERROR: Exceptions caught in Qt event loop:
________________________________________________________________________________
Traceback (most recent call last):
  File "/home/user/src/securedrop-client/securedrop_client/gui/widgets.py", line 2545, in animate_header
    self.header_spinner_label.setPixmap(self.header_animation.currentPixmap())
RuntimeError: wrapped C/C++ object of type QLabel has been deleted
________________________________________________________________________________

[sssoleileraaa]

@emkll curious how many times you had to run make test-random so I can try repro'ing

[sssoleileraaa]

also, could you post your results of the make test runs (i ran 100 times without failure). make test-random I ran 10 times I think and I'm going to increase it now to 20 or more since it takes a long time.

[emkll]

could you post your results of the make test

make test is working fine, I've tried over 20 runs and it's reliable

In sd-app, make test-random is now almost working as expected (had 1 failure due to deleted object in 20 runs)

I'm inclined to merge as-is based on my results described above and the variance in my local environment, as this is already a significant improvement

[emkll]

Merging per the conversation in standup today, thanks @creviera !