Fix JsException passing from pyodide environment to host

[koenvo]

When a JsException exception is raised the pyodide environment pickles the exception and send it back to the host. The host tries to unpickle it and fails. This happens because pyodide.JsException does not exist.

This PR solves this issue by interception during unpickling.

note: maybe a more generic way of passing exceptions from pyodide environment to host is better? cloudpickle might do the trick.

[hoodmane]

Thanks @koenvo! It looks good to me.

[koenvo]

Thanks @koenvo! It looks good to me.

I noticed some tests were failing. Hope all run correctly now.

[rth]

Thanks @koenvo . LGTM as well.

note: maybe a more generic way of passing exceptions from pyodide environment to host is better? cloudpickle might do the trick.

Indeed that's a good question. We could consider switching to cloudpickle if we run into other frequent pickling issues.

[rth]

Could you please just add a changelog.md entry?

[rth]

main / test (macos-latest, 0.21.0, selenium, safari, 1) (pull_request) Failing after

Not sure what's going on with that test, but it's failing quite frequently (unrelated to this PR).

[ryanking13]

main / test (macos-latest, 0.21.0, selenium, safari, 1) (pull_request) Failing after

Not sure what's going on with that test, but it's failing quite frequently (unrelated to this PR).

It seems like safaridriver is a bit unstable. Probably we can disable a test using selenium_standalone_refresh fixture which fails more frequently.