Optimize cargo flaky

Currently, the tests appear to be flaky in our CI (tldr: they're not).

That's because cargo-flaky doesn't handle workspaces correctly. A simple fix is to simply cd in each sub-directory to run flaky test instead of running it at the root of the project. But the real fix should be implemented in cargo-flaky; since our use-case is pretty simple and we're our only users, it should be fairly simple to extract all the existing modules and their associated paths and see if it matches the binary we're trying to run; if it's the case, we execute the process from this path.

In the future, we should definitely look into a better way to handle this process, I'm sure we're not the first one stumbling on this kind of issue. Also, I took a quick glance at cargo-nextest and I must admit I did not really understand when they're changing their path before running their test, so maybe there is another way that I don't know.