Currently the required test coverage is set to a magic number 99%. This is not ideal for at least two reasons:

• The missing "one percent" is still a bit wide, meaning that new code with less than 100% coverage can be submitted, but the coverage check might not catch that.
• One could legitimately refactor some code and removing some (currently covered) lines in the process. As a result, the relative coverage can decrease. If this reduction is just enough to drop under the current 99% threshold, that PR would see a CI failure, even though there's nothing wrong with it.

The generated code has had 100% coverage for quite some time now, meaning that we could clean up the tests a bit and bump the coverage to 100%, avoiding the abovementioned issues.