During the testing gaps analysis, layered cursors were identified as the most undertested area across all foundations-owned components (you can find the full analysis in the comments). They also caused us a lot of trouble during private preview, with most of the private preview blockers being caused by bugs in layered cursors.

The scope of this activity is to revisit the analysis below, identify the most important and risky gaps, determine what we can realistically cover given the time constraints, and move the remaining work into a separate ticket (and maybe to the global testing improvement plan to ensure it does not get forgotten).

One thing to note is that, even if some cases are already covered in test/format, we still believe it is worth adding tests for the same cases in Python or, even better, adding unit tests, since debugging a test/format failure can be very time-consuming and we should not rely solely on this type of testing.