CI testing improvements for run_local.sh

[lkatalin]

It would be easier to read the output and also faster to rerun tests if we could specify a subset of the tests to run.

[lkatalin]

It is also not very clear when tests on the Python agent stop and tests on the Rust agent begin.

[lkatalin]

Does captured stdout refer to the output directly above or below?

[maugustosilva]

@lkatalin @mpeters @THS-on @ansasaki @stefanberger Just wanted to confirm that we plan to keep maintaining run_local.sh (which is broken for both rust and python agents at the moment). My vote is to keep it, but I am open to sunsetting it the majority decides it is not worth the trouble.

[lkatalin]

Would it make sense to drop or rework the tests that seem broken or no longer applicable? It seems good to have a way to run integration tests locally, but those could (and maybe should) look very different from the current run_local.sh.

[stefanberger]

There should be an easy way to run all these tests locally.

When it comes to the rust agent, it should be possible to cache the results, maybe by checking the rust agent out into (hidden) directory of the local keylime checkout because watching it download and build the Rust agent for several minutes every time one runs the local tests isn't very productive.

[mpeters]

I think @ansasaki had some thoughts on why it was broken and some potential fixes. I agree that we should keep it and always have a way of running the tests locally.

[maugustosilva]

I believe PR #1325 basically fixed it. I will close it for now, feel free to open in case you believe it still needs more work.