When using crux-mir with coverage reporting enabled, we often do not want to achieve 100% coverage. For example, we are proving that a panic! call is never reached, or an assert! (either Rust or crucible) is never false.

Currently I would just review the coverage report, and decide which cases are false positives (e.g. a branch with panic! is not called) or true findings (e.g. an if some_setting statement is never true).

The general solution would be to have a mechanism to mark certain functions as "undesirable", and basically don't count paths that lead to such functions.