Moar testing

[vtnate]

DISCUSSION! Running our two_ghe test takes too much memory/cpu for the (free) GHA runner, which cancels the test. Googling around found https://github.com/actions/runner-images/issues/6680 which explains it. This implies our options are:

1. Splash cash for bigger runners
2. Honey I shrunk the test
3. Not run the two_ghe test (at least comment it so it doesn't run automatically on CI)

I think option 3 is the best combination of low cost but still allows us to run the test locally. @mitchute wat do?

Some of the tests weren't running because the paths were outdated. This PR:

• Fixes the paths
  • And DRYed up the paths
• Added some assertions to the network tests
  • And handled the fact that we now re-write the sys-param files in-place
• Revived the GHE test that was commented
• Declares support for Python 3.12
• Adds comments for developer clarity
• Clarifies & updates the readme
• Officially updates version to 0.2.3

[mitchute]

@vtnate my first question is why the two ghe test is consuming so much memory/resources? I'm not necessarily expecting you to have an immediate answer, but GHEDesigner should be pretty capable of handling GHE with several thousand boreholes. If we are attempting GHE that are HUGE, then I say, yes, let's shrink it. If not, there must be some other issue.

[vtnate]

Thanks @mitchute, that's good context. I'll dig into the profiling to see where this is blowing up. GHA doesn't tell me if the problem is memory or cpu, just that it's one of those. Nor does GHA tell me what the limits are, only that we're over them. 🙄