test(fixtures): cache the base example project directory

[codejedi365]

Purpose

Improve speed of test code and reduce the number of IO operations

Rationale

In other development towards changelog testing, I came across the high use of IO operations as we rebuild a project with each test case. This causes a lot of IO operations, and even more so on the use of git. I followed the rabbit hole, and this PR is the first of a few that help increase the test execution speed through deliberate abstraction/modularity. I felt it all at once would be way to confusing and too much for a good review.

How I tested

Besides checking the result here in the CI, I originally was able to confirm the caching mechanism through the VSCode debugger and many break points through the code to see how many times each fixture was called. Once this commit was implemented, the session scope ensured that it was only called once but the init fixture called every test run.

How to Verify

To get VSCode to run the debugger, you have to change the tool.pytest.ini_options.addopts by removing the --cov* options and then changing -nauto to -n0. Once you do this, add a breakpoint to the start of init_example_project, and cached_example_project. Then execute a test (a minimum of 2 that use init_example_project) with the debugger. When you step through, you will see how many times each is called, and you will see the cleanup occur as well if you put a few breakpoints in teardown_cached_dir.

[codejedi365]

@bernardcooke53, pr ready for review for when you have time, thanks!

[bernardcooke53]

I like this a lot @codejedi365, the amount of IO was always quite intense - I'm finding it painfully slow to work on a Mac with the tests on master. Thanks!