Reusing the same image means we don't need to upload so many images to S3 bucket (or multiple buckets if mirrors are configured), significantly speeding up creating larger data sets.
The implementation reuses the path of the first created image, which uses the team-package-version naming pattern, instead of using something more descriptive like "default.png". Bypassing the generated name would have required hacky stuff for the function that calculates the value for ImageField's upload_to attribute, or the models save method. As this would make it theoretically possible to things go wrong in the production, it was deemed better to just use the misleading image name in test environments.
Reusing the same image means we don't need to upload so many images to S3 bucket (or multiple buckets if mirrors are configured), significantly speeding up creating larger data sets.
The implementation reuses the path of the first created image, which uses the team-package-version naming pattern, instead of using something more descriptive like "default.png". Bypassing the generated name would have required hacky stuff for the function that calculates the value for ImageField's upload_to attribute, or the models save method. As this would make it theoretically possible to things go wrong in the production, it was deemed better to just use the misleading image name in test environments.