♻️ Organize polyfills in prep for type-checking and core extraction

[rcebulko]

No description provided.

[rsimha]

@jridgewell @rcebulko Looks like this PR introduced a silent failure in the unit tests.

Two points:

• I've sent out 🏗 Flag an error and exit when Karma runs zero tests #34040 to throw when zero tests are run instead of warning and silently passing. See 🏗 Flag an error and exit when Karma runs zero tests #34040 (comment) for the history on this topic. (A more correct fix would be to get karma-esbuild to propagate its failures to karma. I suspect this isn't happening right now.)
• I will attempt to fix this via a separate PR (since conveniently, I'm build on-duty this week). @rcebulko Let me know if you prefer to do this yourself. Edit: Tests have been fixed by ✅ Fix broken unit tests #34043.

[rcebulko]

Thanks for looking into this. This is the second time we've had something fail tests but not builds; is this #34040 the same issue as before, or a separate one?

[rsimha]

This is the second time we've had something fail tests but not builds;

In this case, the incorrect paths were all in test/unit/polyfills/, so it's expected that building the code (amp dist) would work, but not tests (amp unit). The tests should have failed, but didn't due to the reason described above.

is this #34040 the same issue as before, or a separate one?

Probably the same issue. #34040 fixes it at our end by throwing when zero tests are detected (this should never happen normally). @jridgewell Do you think we can get karma-esbuild to throw when it encounters transformation errors?