[WIP] Add more GUI tests

[robintw]

🧰 Issue

Fixes #988.

🚀 Overview:

🔗 Link to preview (or screenshot, if relevant)

🤔 Reason:

🔨Work carried out:

• Tests pass

🖥️ Screenshot

Confirmations

• I have chosen reviewers for my PR.
• I have chosen an appropriate label for the PR, adding interactive_review if reviewers will need to see UI
• I have extended/updated the documentation in \docs folder
• Any database content changes (Create, Edit, Delete) are recorded in the Log/Changes tables
• Any database schema changes are implemented via alembic revision transitions
• I have completed the mandatory sections of this document.
• I have deleted any unused sections.

📝 Developer Notes:

[codecov]

Codecov Report

Merging #1042 (99bd48f) into develop (c05a0d9) will increase coverage by 0.28%.
The diff coverage is n/a.

❗ Current head 99bd48f differs from pull request most recent head bd23c86. Consider uploading reports for the commit bd23c86 to get more accurate results

@@             Coverage Diff             @@
##           develop    #1042      +/-   ##
===========================================
+ Coverage    78.89%   79.18%   +0.28%     
===========================================
  Files          104      104              
  Lines        10903    10903              
===========================================
+ Hits          8602     8633      +31     
+ Misses        2301     2270      -31     

Impacted Files	Coverage Δ
...epys_admin/maintenance/widgets/custom_text_area.py	63.33% <0.00%> (+1.11%)	⬆️
pepys_admin/maintenance/gui.py	38.88% <0.00%> (+2.69%)	⬆️
pepys_admin/maintenance/widgets/filter_widget.py	53.70% <0.00%> (+6.48%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 61e9be4...bd23c86. Read the comment docs.

[robintw]

As a final update before I leave:

This PR is kinda in a mess right now. There is a useful additional test in tests/gui_tests/test_gui_methods.py that doesn't depend on any of the complex prompt_toolkit stuff. That should probably be merged, as it tests an error-prone part of the GUI implementation.

The tests that rely on interacting with prompt toolkit are broken at the moment. You can see the new tests in test_gui_interactively.py and they are lovely in many ways: written clearly and succinctly, with some nice functions and tooling around them to make them easy to write. Unfortunately they're just very flaky and have particular problems on CI. I've emailed Jonathan (prompt toolkit author) to discuss with him - and will post any response I get here (it should be cc'd to Ian too).