fix(ci): test runs on windows runner

[vados-cosmonic]

This commit fixes an somewhat odd problem with vitest on windows runners in that it cannot process properly await import(...) statements.

The changes in this commit represent the minimal changes to make the tests run properly instead of hanging, though there are other improvements to CI forthcoming.

[vados-cosmonic]

So there was a test failure from a fetch call not resolving but note that the other Windows run (release mode) passed the first time, much better than a stall that never finishes. 🎉

[tschneidereit]

In theory it'd be interesting to see what the tsc compiled code looks like here, and whether that indicates any explanation. In practice though, you've already spent a ton of time on this, and we should just land it and be done with it. Thank you for getting this sorted!

[vados-cosmonic]

Yeah so what's really weird is that theoretically this code shouldn't even be converted by tsc per-say since they're all JS files -- the bug here might be in the transformation at the vite level -- so it might trace to esbuild at the end of the day.

One thing I didn't try was slotting in something like swc instead of esbuild, but it also wasn't easy to find how to even configure vite through vitest -- I guess a vite config file in the right place might do it, maybe.