Most large software projects have something like this. The typical use case is rolling out a feature gradually, to limit the number of users impacted if a new f

Most large software projects have something like this. The typical use case is rolling out a feature gradually, to limit the number of users impacted if a new feature has problems: first you include the feature in a release, but disabled by default; then you turn it on for a small percentage of users. If instrumentation from those users reports crashes, you abort; otherwise, you enable it for everyone else.