In empirical software engineering research, we use "replication package" as term of art for artifacts shared to enable others to rerun a scientific experiment.

Sadly, it is a semantically wrong expression. It should be called "reproducibility package" instead.

The former (replication) is for a "Different team, [with a] different experimental setup" to rerun experiment, which is not the goal.

The latter (reproducibility) is for a "Different team, [with the] same experimental setup", which is.