Improve testing of `save` and `export` functionality

[adamltyson]

The testing of the GUI functionality is a bit hacky. There are sleep() calls, and we don't test the button clicking and the desired functionality separately.

---

Based on comments in #114 by @alessandrofelder:

I think a better long-term solution would be to:

• mock the brainreg_segment.segment.run_save() function in these tests, and check that it's called with the expected parameters
• write unit tests for run_save() and/or export_all()
  • tests for run_save could again mock export_all to avoid those tests being non-deterministic due to multiprocessing
  • tests for export_all could mock the thread_worker decorator to render it trivial and have export_all run in the same thread. (we trust thread_worker is tested sufficiently in the third party library it comes from and so don't need to test it)

This would ensure cleaner separation between

• the tests that check that clicking GUI buttons call the expected functionality
• the tests that call multiprocessing functionality
• the tests that check the multiprocessing functionality itself

[alessandrofelder]

https://pytest-qt.readthedocs.io/en/latest/signals.html might also be useful in this context!