We only run a subset of our tests under GitHub actions, for a variety of reasons. Unfortunately, the GHA tests are the only ones that external contributors can see; evergreen isn't available to them, so if any of the tests we don't run in GHA fail, we have to communicate the failure to the contributors and ask them to effectively troubleshoot it blind. That, or we have to fix it for them.

While it probably isn't reasonable to enable all tests under GHA, we can probably get more coverage than we already do. For example, is it possible to enable the FLE specs there? Are there others we could add?

It's worth an investigation, I think.